epistemology and spiritual authority
DESCRIPTION
The Development of Epistemology and Logic in the Old Nyaya and the Buddhist School Of Epistemology with an annoteted translation of Dharmakirti's Pramana Vartitka II ( Pramana Siddhi ) VV. 1-7 by Vittorio A. Van BiljertTRANSCRIPT
WSTB 20
WIENER STUDIEN
ZUR TIBETOLOGIE UND BUDDHISMUSKUNDE
HERAUSGEGEBEN VON
ERNST STEINKELLNER
HEFT 20
WIEN 1989
ARBEITSKREIS FUR TIBETISCHE UND BUDDHISTISCHE STUDIENUNIVERSITAT WIEN
VITTORIO A. VAN BIJLERT
EPISTEMOLOGY AND SPIRITUAL AUTHORITY
THE DEVELOPMENT OF EPISTEMOLOGY AND LOGIC IN THEOLD NYAYA AND THE BUDDHIST SCHOOL OF EPISTEMOLOGY
WITH AN ANNOTATED TRANSLATION OF DHARMAKIRTI'SPRAMANAVARTTIKA II (PRAMANASIDDHI) W . 1-7
WIEN 1989
ARBEITSKREIS FUR TIBETISCHE UND BUDDHISTISCHE STUDIENUNIVERSITAT WIEN
Preis: OS 230,-
zu beziehen von:
Arbeitskreis fur Tibetische und Buddhistische StudienMaria Theresien-Stra|3e 3/4/26, 1090 Wien, Austria
PREFACE
This book was previously brought out as a PhD-thesis entitled
"The Buddha as a Means of valid Cognition". Originally it had
been my intention to add new findings to this edition, findings
concerning the beginnings of Indian epistemology and logic, such
as are found scattered in various texts antedating the Nyaya
Sutra, as well as further discussion on the interpretation of
Buddhist logic. But certain technical problems and the bulk and
nature of the new material prevented me from adding much to the
present text. They will appear in a separate publication.
Printing errors have been corrected as far as possible, although
I am sure that some have escaped notice, for which I ask the
indulgence of the reader. Certain numberings in the index will at
first sight seem rather erratic, this is due to technical
problems, but if the numbering is read in a specific mathematical
order, the text-place can be easily found.
It remains for me to express my gratitude especially to the
following persons who have contributed greatly to the making of
this book: Dr. T.E. Vetter, Dr. J.C. Heesterman, Leiden, and
Dr. E. Steinkellner, Dr. G. Oberhammer and their pupils, Vienna,
for their scholarly comments on my translations and interpreta-
tion of certain passages; Mrs. Upinder Singh, Delhi, for checking
my English; Mr. T. Meindersma, Groningen, for having been a
willing victim of my classes on Indian epistemology and logic,
the firm Tekstyle, Amsterdam for typesetting this book, and
lastly my grandmother for her encouragement.
The Hague, 1988 V.A. van Bijlert
Table of Contents
Abbreviations
Introduction
Notes
VII
IX
XIX
XXII
CHAPTER I The Beginnings of Systematic
Epistemology and Logic
1.1 The highest Good
1.2 The theory of knowledge, the
four Means of Valid Cognition
1.3 Perception
1.4 Inference
1.5 Comparison based on analogy
1.6 Reliable Statement
1.7 The Syllogism
1.8 Trustworthiness of the Speaker
Notes
CHAPTER II Buddhist Epistemology and Logic
before Dharmakirti
2.1 Vasubandhu
2.2 Perception
2.3 Inference
2.4 The Syllogism
2.5 Dignaga
2.6 The two objects of knowledge
2.7 Perception
2.8 Pramana and its
Result are not different
2.9 Inference for oneself
2.10 The syllogism
2.11 Reliable statement
Notes
11
4
7
10
15
16
20
30
35
45
45
45
48
50
54
55
58
62
64
70
80
83
VIII
CHAPTER III Dharmakirti's Logic 93
3.1 The utility of Inference 93
3.2 A compendious definition
of the Probans 94
3.3 Effect as Probans 99
3.4 Essential property as Probans 101
3.5 Non-perception as Probans 102
3.6 Four kinds of Non-perception 104
3.7 The DrstSnta 107
Notes 111
CHAPTER IV The Pramana-definitions of the
Pramlfnasiddhi chapter of PV, vv 1-7 115
lab 120
lcb 125
led 130
2 132
3ab 140
2bd 141
4ac 146
4d-5a 147
5b 149
5c 150
5d-6a 153
6bc 155
Index 181
IX
Abbreviations
AKBh Abhidharmakosabhasyam of Vasubandhu, ed.
by P. Pradhan, Patna 1975, Tibetan
Sanskrit Works Series vol. VIII
Burrow 1980 Burrow, T. , Sanskrit Ma- 'to ascertain1,
Transactions of the Philological Society,
Oxford 1980, p. 135-140
Chakrabarti 1977 Chakrabarti, Kisor Kumar, The Logic of
Gotama, The University Press of Hawaii,
19 77, Monograph No. 5 of the Society for
Asian and Comparative Philosophy
Chemparathy 1983 Chemparathy, G. , l'Autorite du Veda selon
les Nyaya-Vaisesikas, Louvain-La-Neuve
1983
- Derge edition
DhP Dharmottarapradipa of Durvekamisra,
being a subcommentary on Dharmottara's
Nyayabindutlka, ed. by Dalsukhbhai
Malvania, Patna 19712, Tibetan Sanskrit
Works Series, vol. II
DvNC II Dvadasaram Nayacakram of Acarya SrT
Mallavadi Ksamasramana with the commentary
Nyayagamanusarini of Sri Simhasuri Gani
Vadi Ksamasramana; part II, ed. by Muni
Jambuvijaya-jT, Sri Jain Atmanand
Sabha-Bhavnagar, 1976, §ri Atmananda Jain
Granthamala Serial No. 94
Frauwallner 1933 Frauwallner, E., Zu den Fragmenten
buddhistischer Logiker im Nyayavarttikam,
WZKM 40, 1933, p. 281-304
Frauwallner 1935 Frauwallner, E. , Beitrage zur Apohalehre,
WZKM 42, 1935, p. 93-102
Frauwallner 1954 Frauwallner, E. , Die Reihenfolge und
Entstehung der Werke Dharmaklrti's,
Asiatica, Festschrift F. Weller, Leipzig
1954, p. 142-154
Frauwallner 1957 Frauwallner, E. , Vasubandhu's Vadavidhih
WZKSO 1, 1957, p. 104-146
Frauwallner 1957a Frauwallner, E. , Review of the edition of
PVBh in JAOS 77, 1957, p. 58-60
Frauwallner 1959 Frauwallner, E. , Dignaga, sein Werk und
seine Entwicklung, WZKSO 3, 1959,
p. 83-164
Frauwallner 1960 Frauwallner, E., Devendrabuddhi, WZKSO 4,
1960, p. 119-123
Frauwallner 1961 Frauwallner, E. , Landmarks in the History
of Indian Logic, WZKSO 5, 1961, p. 125-148
Frauwallner Gesch II - Frauwallner, E. , Geschichte der Indischen
Philosophie, II. Band, Otto Mtiller Verlag,
Salzburg 1956
Goekoop 19 67 The Logic of the invariable concommitance
in the Tattvacintamani, Reidel, Dordrecht
1967
GOS - Gaekwad Oriental Series
Gupta 1962 Gupta, B. , Die Wahrnehmungslehre in der
Nyayamanjarl, Bonn 1962
XI
Hattori 1968 Dignaga, On Perception, being the
Pratyaksapariccheda of Dignaga's
Pramanasamuccaya, transl. and ann. by
Masaaki Hattori, Harvard University Press,
Cambridge, Massachusetts, 1968
Hayes 1980 Hayes, Richard P., Difinaga' s Views on
Reasoning (Svarthanumlna), JlPh 8, 1980,
p. 219-277
IIJ - Indo-Iranian Journal
JAOS - Journal of the American Oriental Society
JBORS The Journal of the Bihar and Orissa
Research Society
JlPh
Katsura 1975
Katsura 1984
- Journal of Indian Philosophy
- Katsura, Shoryu, New Sanskrit Fragments of
the PramSnasamuccaya, JlPh 3, 1975,
p. 67-78
- Katsura, Shoryu, Dharmaklrti's theory of
truth, JlPh 12, 1984, p. 215-235
Kitagawa 19 65 Kitagawa, Hidenori, Indo koten ronrigaku
no kenkyu: Jinna (Dignaga) no taikei,
Kyoto 1965 (The book contains an edition
of the Tibetan translations of PS,
Sanskrit fragments, and a Japanese
translation.)
Kneale Kneale, W. and Kneale, M., The Development
of Logic, Clarendon Press. Oxford 1978
XII
Lindtner 1980 Lindtner, Christian, Apropos Dharmakirti -
Two new Works and a new Date, Acta
Orientalia 41, 1980, p. 27-37
Nagatomi 1959 Nagatomi, M. , The Framework of the
Pramanavarttika Book I, JAOS 79, 1959,
p. 263-266
Nagatomi 1967/1968 Nagatomi, M. , Arthakriya, The Adyar
Library Bulletin, No. 31-32, ly67-1968,
p. 52-72
NB - see DhP
NBh The Nyayadarsana of Gotama with the BhSsya
of Vatsyayana, ed. by Padmaprasada Sa"strl
and Harirama Sukla, Varanasi 19833, Kashi
Sanskrit Series 43
ND Nyayadarsanam, with Vatsyayana fs BhcEsya,
Uddyotakara's Varttika, Vacaspati Misra's
Tatparyatlka and Visva-natha1s Vrtti, ed.
by Taranatha Nyaya Tarkatirtha and
Amarendramohan Tarkatirtha, Calcutta
1936-1944, Reprint Rinsen Sanskrit Text
Series, 1-2, Kyoto 1982
NM - Nyayamukha of Dignaga
NS - Nyaya Sutra, see NBh
NS (G) Ganganatha Jha, M., The Nyaya-Sutras of
Gautama with Vatsyayanafs Bhasya and
Uddyotakara1s Varttika, Allahabad 1915,
Indian Thought Series, No. 7, vol. I-IV
XIII
NS (R) Die Nyayasutra1s, Text, Ubersetzung,
Erlauterungen und Glossar von Walter
Ruben, Leipzig 1928, Kraus Reprint Ltd.
Nendeln, Liechtenstein 1966
NS (V) The Nyaya Sutras of Gotama, translated by
M.M. Satisa Chandra Vidyabhushana, revised
and ed. by Nandalal Sinha, Motilal
Banarsidass, Delhi 1981
NV Nyayavarttika of Uddyotakara, Kashi
Sanskrit Series No. 33, Benares 1915-1916
OAdW Osterreichische Akademie der Wissen-
schaften
Oberhammer 1963
Oberhammer 1964
Oberhammer 1964a
Oberhammer, G. , Ein Beitrag zu den
Vada-Traditionen Indiens, WZKSO 7, 1963,
p. 63-103
Oberhammer, G. , Der Svabhavika-sambandha,
ein Geschichtlicher Beitrag zur Nyaya-
Logik, WZKSO 8, 1964, p. 131-181
Oberhammer, G. , Paksilasvamin1s Intro-
duction to his Nyayabhasyam, Asian
Studies, vol. II, No. 3, Dec. 1964,
University of the Philippines, p. 302-322
Oberhammer 1966 Oberhammer, G., Zur Deutung von Nyaya-
sutram 1.1.5, WZKSO 10, 1966, p. 66-72
Oberhammer 1974 Oberhammer, G. , Die Uberlieferungs-
autoritSt im Hinduismus, in: Offenbarung,
Geistige RealitSt des Menschen, Wien 1974,
p. 41-92
XIV
P - The Tibetan Tripitaka, Peking Edition, ed.
by D.T. Suzuki, 168 vols, Otani Univer-
sity, Tokyo-Kyoto 1955-1961
Potter 1977 - Potter, Karl H. (ed.), Encyclopedia of
Indian Philosophies, vol. II, Nyaya-
Vai§esika, Motilal Banarsidass, Delhi 1977
PS - Pramanasamuccaya
PS I - see Hattori 1968
PS II - see Hayes 1980
PV - Pramanavarttika
PV I - The Pramanavarttikam of Dharmakirti, The
first chapter with the Autocommentary, ed.
by Raniero Gnoli, Roma 1960, Serie
Orientale Roma 23
PV II - Pramanavarttika II, Pramanasiddhi
PV I (Mi) - Pramanavarttika-karika (Sanskrit and
Tibetan), ed. by Yusho Miyasaka, Acta
Indologica II, Naritasan Shinshoji
1971/1972, p. 2-41
PV II trsl Vetter - PV II. 131-285, Der Buddha und seine Lehre
in Dharmaklrtis Pramanavarttika, einge-
leitet, ediert und tibersetzt von Tilmann
Vetter, Arbeitskreis flir Tibetische und
Buddhistische Studien UniversitSt Wien,
Wien 1984, Wiener Studien zur Tibetologie
und Buddhismuskunde, Heft 12
XV
PV III
PVBh
- Pramanavarttika III, Pratyaksapariccheda
- Prajnakaragupta1s Pramanavarttikabhasya
and Varttikalankara, ed. by R.
Samkrtyayana, Kashi Prasad Jayaswal
Research Institute, Patna 1953
PVin
PVP
- Pramanaviniscaya
- Pramanava'rttikapanjika: P 5717
(b), Tshad ma, Che, l-390a.8; D4217, Tshad
ma, Che lb-326b
PVSV - Pramanavarttikasvavrtti
PVSV (Mai)
PVSVT
Pramanavarttika, Svarthanumana-Pariccheda
by Dharmakirti, ed. by Dalsukh Bhai
Malvaniya, Varanasi 1959, Hindu
Vishvavidyalaya Nepal Rajya Sanskrit
Series, vol. 2
Acarya DharmakTrteh Pramanavarttikam
(Svarthanumanaparicchedah) svopajnavrttya,
Karnakagomiviracitaya tattikaya ca
sahitam, ed. by Rahula Samkrtyayana, Kitab
Mahal, Allahabad 1943; Reprint: Karnaka-
gomin's Commentary on the Pramana-
varttikavrtti of Dharmakrrti, Rinsen Book
Co., Kyoto 1982, Rinsen Buddhist Text
Series I
P W Pramanavarttikavrtti: Dharmaklrti's Pramana-
varttika with a commentary by Manorathanan-
din, ed. by R. Samkrtyayana, JBORS 24-26
XVI
PW(S) Pramanavlrttika of Acharya Dharmaklrti
with the commentary 'Vritti1 of Acharya
Manorathanandin, crit. ed. by Swami
Dvarikadas Shastri, Varanasi 1968, Bauddha
Bharati Series 3
PVT Pramanavarttikatika Tib. translation in P
5718, Tshad ma, Je, l-402a.8; Ne,
l-348a.8; PVT I, Je lb.1-Ne 85b.2, PVT II,
Ne 85b.2-Se 187a.8
Randle
Randle Frag
Randle, H.N. , Indian Logic in the Early
Schools, Humphrey Milford, Oxford Univer-
sity Press London, New York, Bombay,
Calcutta, Madras 1930, Indian Reprint,
Motilal Banarsidass, Delhi 1976
Randle, H.N., Fragments from Dinnaga,
Indian Reprint, Motilal Banarsidass, Delhi
1981
Ruben
Steinkellner 1961
Steinkellner 1966
Steinkellner 1967
Steinkellner 1971
- see NS (R)
- Steinkellner, E. , Die Literatur des
Alteren Nyaya, WZKSO 5, 1961, p. 149-163
- Steinkellner, E. , Bemerkungen zu
Tsvarasenas Lehre vom Grund, WZKSO 10,
1966, p. 73-85
- Steinkellner, E., Dharmaklrtifs Hetubinduh
I-II, Teil I Tibetischer Text und rekon-
struierter Sanskrit-Text, Teil II Uber-
setzung und Anmerkungen, OAdW, Wien 1967
- Steinkellner, E., Wirklichkeit und Begriff
bei Dharmaklrti, WZKS 15, 1971, p. 179-211
XVII
Steinkellner 1973 Steinkellner, E. , Dharmaklrti's Pramana-
viniscayah, Zweites Kapitel: Svarthanu-
manam, Teil I, Tibetischer Text und
Sanskrittexte, OAdW, Wien 1973
Steinkellner 1974 Steinkellner, E., On the Interpretation of
the Svabhavahetuh, WZKS 18, 1974,
p. 117-129
Steinkellner 1979 Steinkellner, E. , Dharmaklrti's Pramana-
viniscayah, Zweites Kapitel: Svarthanu-
manam, Teil II, Ubersetzung und Anmer-
kungen, OAdW, Wien 1979
Steinkellner 1981 Steinkellner, E. , Philological Remarks on• _
Sakyamati's Pramanavarttikatika, in:
Studien zum Jainismus und Buddhismus,
Gedenkschrift fur Ludwig Alsdorf, Franz
Steiner Verlag, Wiesbaden 1981, p. 283-295
Steinkellner 1982 Steinkellner, E. , The Spiritual Place of
the Epistemological Tradition in Buddhism,
Nanto Bukkyo 49, 1982, p. 1-18
Tillemans 1984 Tillemans, Tom. J.F., Sur le Parar-
thanumana en Logique Bouddhique, Asia-
tische Studien / Etudes Asiatiques, 38.2,
1984, p. 73-99
TTP - see P
Tucci 1929 Tucci, G. , Pre-Dignaga Buddhist Texts on
Logic from Chinese Sources, Baroda 1929,
GOS
XVIII
Tucci 1930 Tucci, G., The Nyayamukha of Dignaga, The
oldest Buddhist Text on Logic, after
Chinese and Tibetan Materials, Heidelberg
1930, Materialien zur Kunde des
Buddhismus, 15. Heft
Vetter 1964 Vetter, Tilmann, Erkenntnisprobleme bei
Dharmaklrti, OAdW, Wien 1964
Vetter 1984
Vip
- see PV II trsl Vetter
- Vadanyayaprakarana of Dharmaklrti with the
commentary Vipancitartha of Santaraksita,
crit. ed. by Dwarikadas Shastri, Varanasi
1972, Bauddha Bharati Series 8
Wezler 1968 Wezler, A., Die "Dreifache" Schlussfol-
gerung im Nyayasutra 1.1.5, IIJ 11,
1968-1969, p. 109-211
Wezler 1979 Wezler, A., Dignaga's Kritik an der
Schlusslehre des Nyaya und die Deutung von
Nyayasutra 1.1.5, ZDMG, Suppl. 1.3, 1969,
p. 836-842
WZKM Wiener Zeitschrift fur die Kunde des
Morgenlandes
WZKS - Wiener Zeitschrift fur die Kunde Stidasiens
WZKSO Wiener Zeitschrift fiir die Kunde Slid- und
Ostasiens
ZDMG Zeitschrift der Deutschen MorgenlMndischen
Gesellschaft
XIX
Introduction
In this book I present an interpretative translation with ex-
haustive commentary of Dharmaklrtifs Pramanavarttika (PV)
II.1-7 . These seven verses put forward and briefly discuss two
general definitions of the notion 'pramana', 'means of valid
cognition1, the most basic notion in old Indian epistemology.
Before Dharmaklrti wrote these verses, no general pramana-
definitions had been given in any of the older epistemological
treatises. Therefore Pramanavarttika II.1-7 occupies a rather
unique place in Indian epistemology.
The two definitions are meant to provide a sound epistemolo-
gical basis for a long discussion in the rest of PV II on the
specific causes and characteristics of the religious authorita-
tiveness of the Buddha and the most important aspects of His
doctrine, as well as the possibility and the boundaries of such
an authoritativeness .
The conciseness of Dharmaklrti's style of writing, especially
in PV II.1-7, forced me to provide the translation with supple-
mentary words, phrases or even whole sentences between square4
brackets in order to make for intelligible reading, as a mere
word-for-word rendering would hardly convey any meaning. This
method I have used throughout the book. Apart from PV II.1-7
itself, I have translated (for the first time in a European
language) the relevant exegetic passages from the oldest
commentary on PV II-IV, the Pramanavarttika-Panjika (PVP), of
Dharmaklrti's direct pupil Devendrabuddhi . My representation of
what is taught in PV II.1-7 is thus supported and verified by a
commentary which indigenous tradition explicitly or implicitly
has always regarded as authoritative. In explaining PV II.1-7 I
have also drawn on Dharmaklrtifs first work, the Pramanavarttika
(PV) I, together with the Pramanavarttika-Svavrtti (PVSV), in
which he has formulated for the first time his new system of
logic, his philosophical semantics, and his ideas on the relia-
bility of language as a means to communicate correct knowledge.
Some notions in PV II.1-7 originated from DharmakXrti*s philo-
sophical predecessors among whom Paksilasvamin and Dignaga stand
out because of their original and fundamental contributions to
epistemological and logical thought.
XX
In my commentary on PV II.1-7 I quote important passages from
these authors in order to show the extent to which Dharmaklrti
has used their ideas in this own doctrine.
In the first chapter I describe the epistemology and logic of
the Nyaya-Sutra (NS) and the Nyaya-Bhasya (NBh) . The NS is one
of the earliest purely philosophical treatises which consistently
defines its basic concepts and makes the discussion on the
sources of correct knowledge its prime concern. It is also the
source of a specific scheme of inferential reasoning that later
culminated in Dignaga's system of logic. The NS has to a great
extent provided the terminology and theoretical framework of
subsequent epistemology, logic and dialectics. The NBh discusses
(for the first time) the possibility of acquiring correct know-
ledge, its verifiability and its practical use. Furthermore, it
propounds a logic more refined than that of the NS and submits
(also for the first time) some epistemological criteria to verify
metaphysical statements.
My second chapter shows how the immediate predecessors of^ 8
Dharmakirti, the Buddhist philosophers Vasubandhu and Dignaga,
improved on the epistemology and logic of the NBh. Vasubandhu
wrote two small treatises on epistemology, logic and dialectics,
the Vadavidhana and the Vadavidhi. In these works he offers a
theory of perception that is slightly different from that of the
NBh, but more importantly, he creates a completely coherent
system of logic by drawing the necessary conclusions from some
logical doctrines that had remained scattered and unconnected in
the NBh. Vasubandhu1 s system is described in the first part of
chapter II. The second part is devoted to Vasubandhu1s philoso-
phical successor Dignaga. Dignaga summed up his own theories on
epistemology, logic, dialectics and philosophical semantics in the
Pramanasamuccaya (PS), a handbook which became the main source of
advanced epistemology and logic for Buddhist as well as non-
Buddhist philosophers until it was eventually superseded by
Dharmaklrti"s works. I describe Dignaga's epistemology and logic
almost exclusively by quoting relevant passages from the PS. In
relation to his logic, I also quote from an earlier work of his.
In this way the reader can get a clear picture of (a) the
development from Vasubandhu to Dignaga and
XXI
(b) he has before him those doctrines of Dignaga on which
DharmakTrti has built his own system.
In chapter III I present an outline of Dharmakirti's first
attempt at formulating a new and effectual system of logic.
Taking Dignagafs logic as his starting-point, DharmakTrti greatly
simplifies its inferential procedure and shows how the logical
connexion of concepts must have its basis in reality.
Dharmakirti had not yet formulated his own theory of percep-
tion (the first pramana in Dignagafs as well as his own episte-
mology) when he wrote PV I and PV II. He discusses perception at
length in PV III, which was composed after PV II. Therefore it is
very likely that Dharmakirti, while putting forward his pramana-Q
definitions in PV II. 1-7 , still used as his main point of
departure what Dignaga had said about perception as pramana in PS
I. Hence I have not devoted a separate chapter to Dharmakirti's
perception-theory, but in commenting on the perceptional aspect
of the pramana-definitions in PV II.1-7 I am often referring to
the relevant passages dealing with perception in PS I.
The first three chapters as a whole should be regarded as a
rather lengthy general introduction to the translation of PV
II.1-7. Chapter II and III should be regarded as a commentary on
the notion pramana as it is defined in PV II.lab. For this reason
I do not elaborately explain this word again in the actual
commentary in chapter IV which contains the translation and
interpretation of PV II.1-7.
Untill now, there has been only one publication that exclu-
sively deals with the pramana-definitions and discussions of PV
II.1-7 , namely the article by Sh. Katsura, "Dharmakirti•s
Theory of Truth", which appeared in 1984. In this article Katsura
presents an interpretative translation of PV II.1-6 with a
philosophical commentary of his own in which he explains the two
definitions and summarizes DharmakTrtifs ontological and episte-
mological views. The article is a useful general introduction to
the epistemological problems PV II.1-6 raises. Still, it does not
make my representation of PV II.1-7 in this book superfluous, as
I try to describe those epistemological and logical developments
that have led to Dharmakirti•s theory, translate parts of the
oldest commentary on PV II and show how Dharmakirti uses his two
pramana-definitions to lay down the epistemological and religious
authority of the Buddha.
XXII
NOTES
1. The famous Buddhist philosopher DharmakTrti lived 600-660 AD
according to Frauwallner 1961, p. 137-139, but Lindtner 1980
proposes ca. 530-600 AD. Lindtner's proposal, which is not
universally accepted, rests on the assumption that
DharmakTrti was the author of a Tattvaniskarsa from which
some verses are quoted in Bhavya's Madhyamakaratnapradipa.
2. This chapter is also called Pramanasiddhi, 'The establishing
of (the Buddha as) a means of valid cognition1. About the
order in which the four chapters of PV were written cf
Frauwallner 1954, p. 143, 145, 148-149. This order of the
chapters was kept by the old commentators (the earliest of
whom is Devendrabuddhi). The late commentator Manorathanandin
changed the received order in the following way: chapter I
becomes III, II becomes I, I LI becomes II, IV remains IV.
Manorathanandin1s order is retained, curiously enough, in
Miyasaka's edition of the PV, PV (Mi). In the present book
the original pre-Manorathanandin order is kept.
3. I refrain from giving a detailed exposition as Vetter 1984,
p. 13-35 already deals with the contents of PV II in a very
readable and thoroughgoing manner.
4. This procedure of interpretative translation is adopted by
Vetter in his translation of PV II.121c-279 (PV II trsl
Vetter).
5. According to Frauwallner 1961, p. 145 Devendrabuddhi lived
ca. 630.689 AD.
6. According to Frauwallner Gesch II, p. 22 and Steinkellner
1961, p. 151 the Brahmanic Nyaya philosopher Paksilasvamin
could have lived in the 5th cent AD. He is the author of the
influential Nyaya-Bhasya (NBh), the earliest commentary on
the Nyaya-sutra (NS) that has survived. The Buddhist philo-
sopher Dignaga who lived ca. 540 AD (according to Frauwallner
1961, p. 137) wrote quite a number of works (cf Frauwallner
XXIII
1959, p. 83, 139-164). His most famous book is the Pramana-
samuccaya (PS).
7. The Nyaya-Sutra is the oldest systematic handbook on epis-
temology and metaphysics (cf Ruben, p. XV-XVI) and was
written probably in the 1st or 2nd cent AD (cf Tucci 1929,
p. XXIIIf and Potter 1977, p. 220).
8. This Vasubandhu should have lived ca. 400-480 AD (Frauwallner
1961, p. 129). Thus, he is either a bit later than, or
contemporary with, Paksilasvamin.
9. In the present book, only the first seven verses of PV II are
being translated and discussed. A second publication will
contain the translation and interpretation of PV II.8-34
which contains, among other things, a refutation of various
proofs of the existence of a supreme Creator-God. DharmakXrti
refutes these proofs with the help of his own system of logic
as put forward in PV I and the PVSV.
10. The epistemological and logical English terminology that is
used throughout this book to translate the Sanskrit terms
derives mostly from the following works: S. Chandra
Vidyabhusana, A History of Indian Logic, Calcutta 1920,
reprint Delhi 1979; H.N. Randle, Fragments from Dignaga,
London 1926, reprint Delhi 1981; G. Tucci, The Nyayamukha of
Dignaga, Heidelberg 1930; S. Mookerjee and H. Nagasaki, The
Pramanavarttikam of DharmakTrti, Nalanda, Patna 1964;
C. Goekoop, The Logic of Invariable Concomitance in the
Tattvacintamani, Reidel, Dordrecht 1967; M. Hattori Dignaga,
On Perception, Harvard 1968; E. Steinkellner, On the Inter-
pretation of the Svabhavahetuh, WZKS 18, 1974, p. 117-129.
11. Before Katsura 1984, we find a few valuable remarks on this
passage in Vetter 1964, p. 27-28, 31-32, Steinkellner 1982,
p. 10-13 and there is the Japanese translation of the whole
PV II by T. Kimura, Dharmaklrti ShukyS Tetsugaku no Genten
Kenkyu, Tokyo 1981 (but cf also the remark on it in Vetter
1984, p. 7).
I The Beginnings of Systematic Epistemology and Logic
1.1 The highest Good
The first Indian school of thinking which explicitly
teaches that the supreme good, defined as emancipation
from all suffering, can be the object of rational inquiry
and which, moreover, propounds a system of epistemology,
logic and dialectics in order to give definite shape to
this rational inquiry, seems to be the old Nyaya-school .
The oldest written document of that school is the Nyaya
Sutra (NS), a text consisting of some 530 aphoristic sen-
tences (sutras) arranged in 5 chapters, in which the basic
concepts of the school are defined and discussed. Tradi-
tion has it that the text was written or edited by the
somewhat mythical figure Aksapada Gotama, but there is ho
certainty concerning this personality or when he lived .
Although the NS combines heterogeneous philosophical no-
tions such as ultimate liberation (in the religious sense)
with natural philosophy, epistemology, logic and dialec-
tics, we find in this text a much stronger tendency to-
wards systematic exposition and coherency than in the
philosophical writings that preceded it .
Even though the NS was a more consistent philosophi-
cal work than previous * works and created the conceptual
framework of much of subsequent epistemological and logi-
cal thinking, definite form to the old Nyaya-philosophy
was given by Paksilasvamin Vatsyayana in his Nyaya Bhasya,
the earliest extant commentary on the NS. We do not ex-
actly know when Paksilasvamin lived or where, but it must
necessarily be after the date of composition of the NS and
some time before Vasubandhu, one of the first known Bud-
dhist logicians who improved upon the epistemology, logic
and dialectics of the NS and the NBh4. On the authority of
Frauwallner, we believe Vasubandhu the logician to have
lived between around 400-480 AD (cf Frauwallner 1961, p.
129-131); thus Paksilasvamin wrote his commentary any time
before 480 AD (if we allow for the possibility of these
two having been contemporaries) . Paksilasvamin not only
quite accurately interprets the NS-aphorisms themselves,
he also sets forth many important new theories in addition
to some seminal ideas found in the aphorisms he is explain-
ing. The NS by its systematic framework and the NBh by its
formulating some fundamental epistemological and logical
principles for the first time, gave the impetus to the
Buddhists (Vasubandhu, Dignaga and even to some extent
Dharmaklrti) to develop their own form of "Nyaya", their
own systematic treatment of epistemology, logic and dia-
lectics (cf Steinkellner 1961, p. 150-152).
True knowledge (tattvajnana) is of great importance
to the adherent of the Nyaya-philosophy and hence the
school lays a great deal of stress on epistemology and
logic. However, true knowledge is not sought for its own
sake; it is valued as a necessary condition for acquiring
the supreme good in life6. We find this in NS 1.1.1, an
aphorism which at first sight simply reads like a table of
contents:
pramanaprameyasamsayaprayojanadrstantasiddhantavayavatar-
kanirnayavadaj alpavitandahetvabhasacchalaj atinigraha-
sthananam tattvajnanan nihsreyasadhiqamah,
'The obtainment of the supreme good [proceeds] from the
knowledge of the truth [ie true knowledge] of [ie concern-
ing] (1) the means of valid cognition, (2) the objects
[deserving] to be cognized [through those means of valid
cognition], (3) doubt, (4) motive, (5) generally accepted
fact, (6) established tenet, (7) the members [of the syl-
logism], (8) reasoning, (9) ascertainment, (10) discus-
sion, (11) sophistry, (12) [capricious] faultfinding, (13)
fallacious [logical] reasons, (14) deliberate misunder-
standing, (15) futile rejoinders and (16) grounds for
losing [an argument]1. These sixteen topics as a whole
define the specific nature of the Nyaya-philosophy as
mainly consisting of epistemology, logic and dialectics.
The true knowledge of this philosophy enables one to at-
tain (adhigama) to the highest good (nihsreyasa), after-
wards defined as complete freedom from suffering; it is
said in the NS 1.1.22, tadatyantavimokgo 'pavargah, 'The
complete liberation from that [ie from suffering] is
[final] emancipation17. The ablative case tattvajnanat in
NS 1.1.1 suggests that true knowledge is not the instru-
ment itself by which the highest good is obtained, for in
that case we expect the instrumental case tattvajnanena,
but that true knowledge is an indispensable prerequisite
for obtaining the highest good, so that the good really
proceeds from the knowledge8. The next sutra describes how
the highest good as final emancipation arises from true
knowledge:
duhkhaj anmapravrttidosamithyaj nananam uttarottarapaye
tadanantarapayad apavargah
(NS 1.1.2),'When, of suffering, birth, activity, moral
flaws and false knowledge, every next [item which is men-
tioned here] disappears, then it [ie every item which is
mentioned] immediately [before, and which is the result of
every next item] disappears [and] thereby emancipation
[is won]'.
What is meant in this aphorism is that these five
should disappear in the reverse order. Paksilasvamin makes
this quite clear in his commentary:
yada tu tattvajnanan mithyajnanam apaiti# tada mithya-
jnanapaye dosa apayanti, dosapaye pravrttir apaiti
pravrttyapaye janmapaiti lanmapaye duhkham apaiti,
duhkhapaye catyantiko 'pavargo nihsreyasam iti,
'When on account of knowledge of the truth [ie true know-
ledge] false knowledge disappears, then at the disap-
pearance of false knowledge moral flaws disappear; at the
disappearance of moral flaws activity disappears, at the
disappearance of activity [a new] birth disappears; at the
disappearance of [a new] birth suffering disappears; and
at the disappearance of suffering everlasting emancipa-
tion, the supreme good [is won] (NBh 20.3-5)'. Since false
knowledge indirectly causes suffering — according to NS
1.1.2 and the commentary thereon — , it follows that true
knowledge, being the contrary of false knowledge, ulti-
mately ends all suffering, as is maintained in the NBh-9passage .
Although the source of tattvainana is not specified
in the NS, it is necessarily so that true knowledge ori-
ginates from a source which can be nothing else but the
means of valid cognition, the pramanas. For true knowledge
of a thing, true cognition of reality, is acquired — as
will be explained in the sequel -- through the means of
valid cognition10, a term which renders the Sanskrit word1pramana', also meaning 'measure1, 'standard1, 'criterion1
and 'authority'11 .
1.2 Theory of Knowledge, the four means of valid cognition
Before describing the epistemology and logic of the NS, I
will quote the statements made by Paksilasvamin on the
function and advantages of correct knowledge. In the open-
ing phrase of the NBh he says:
pramanato 'rthapratipattau pravrttisamarthyad arthavat
pramanam,
'A means of valid cognition possesses the object, because
of the appropriateness of [human] activity [with regard to
the object] when the object is cognized through a means of
valid cognition (NBh 1.5) *. The correct knowledge brought
forth by a means of valid cognition enables the cognizer
to act with regard to objects, and enables him (or her) to
make the appropriate choices concerning these objects. The
veracity of the cognition is gauged precisely from this
ability to make the right choices; practical utility is
thus the measuring-rod of truth. The NBh continues:
pramanam antarena narthapratipattih, narthapratipattim
antarena pravrttisamarthyam. pramanena khalv ayam jnata-
rtham upalabhya tarn artham abhipsati jihasati va. tasye-
psajihasaprayuktasya samlha pravrttir ucyate. samarthyam
punar asyah phalenabhisambandhah. samlhamanas tarn artham
abhipsan jihasan va tarn artham apnoti jahati va*. arthas tu
sukhafl sukhahetus ca, duhkhaifi duhkhahetus ca. so 'yam
pramanartho 'parisaiftkhyeyah pranabhrdbhedasyaparisam-
khyeyatvat.
'Without a means of valid cognition there is no cognition
of an object, without cognition of an object there is no
appropriateness of [human] activity. For, when a cognizer
has cognized an object with a means of valid cognition, he
either wishes to obtain that object or to avoid it. The
striving of him who is [thus] instigated by the wish to
obtain or to avoid is called "activity". And furthermore,
the appropriateness is this [activity1s] connexion with a
result. He who is striving, being desirous of obtaining or
avoiding an object, [finally] obtains that object or
avoids it. As to the object, it is happiness and a cause
of happiness, as well as suffering and a cause of suffer-
ing. And the [number of] objectfs] for the means of valid
cognition is limitless, owing to the fact that there are
innumerable kinds of sentient beings (NBh 2.1-3.1)'. Cor-
rect knowledge has practical value inasmuch as it permits
us to choose between what gives happiness and what will
cause grief. What is to be abandoned is that which gives
us pain and misery. With true knowledge we can get to know
the causes of pain and remove them.
Paksilasvamin points in the foregoing statements to a
more secular, practical use of the Nyaya, a use nowhere
explicitly mentioned in the NS. For in the NS the only
good seems to be liberation from suffering in the highest
religious sense (cf NS 1.1.1-2). Paksilasvamin on the
other hand, without in the least detracting from the ul-
timate religious goal, explicitly shows how the Nyaya
philosophy can also help us in making the proper choices
in daily life. The universal applicability of the Nyaya is
implied in the assertion that 'the [number of] object[s]
for the means of valid cognition is limitless, owing to
the fact that there are innumerable kinds of sentient
beings (NBh 3.1-2)'. And we find elsewhere in the NBh:
yam artham abhlpsan j ihasan va karmarabhate tenanena sarve
praninah sarvani karmani sarvas ca vidya vyaptah
tadasrayas ca nyayah pravartate,
'The object which one wishes to obtain or to avoid [and
for the sake of which] one undertakes action, that [very
object] pervades all sentient beings, all actions and all
[forms of] learnedness. And the Nyaya [as system] operates
with that [aforementioned object one wishes to obtain or
avoid] as its basi^ (NBh 8.1-3)f. All activity and learn-
ing eventually aim at procuring the desirable and shunning
the undesirable.
The utility of correct knowledge being in this way
established, we will now have to elaborate on the produc-
tion of knowledge by the four means of valid cognition. NS
1.1.3 enumerates them:
pratyaksanumanopamanasabdah pramanani,fThe means of valid cognition are: perception, inference,
comparison [based on analogy] and [reliable] statement'.
This implies that these four instruments produce all the
knowledge we can possibly acquire. Perception here pri-
marily means sensory perception, ie direct perception by
the eyes, ears, nose, tongue and touch. Inference is a
judgement following on direct perception and revealing ob-
jects indirectly, objects that are not perceived at the
time of inferring. Paksilasvamin glosses:
anumanain, mitena linqenarthasya pascanmanam anumanam,
'inference: the subsequent cognition of an object, with
the help of a property [of that object] which is [already]
known [through a means of valid cognition], is an inferen-
ce (NBh 10.18)'. Thus the internal process of deductive
reasoning as it happens in one's mind is briefly descri-
bed. To give one famous example: if someone sees smoke on
a mountaintop, he can infer that there is fire on the
mountain, for he knows that wherever there happens to be
smoke there is fire. He does not see the fire directly on
the mountain, he 'sees1 it indirectly through the coin-
cidental smoke. Hence it is a cognition subsequent to
(pascad) another cognition. Comparison means to know a new
thing through its similarity to a given thing, a simila-
rity which was previously described by a person who saw
the new thing himself.
upamanam sarupyajnanam yatha gaur evaifl gavaya iti,
Comparison [based on analogy] is knowledge of the sameness
of form: "a bos Gavaeus is like a cow", (NBh 24.2)'. Reli-
able statement gives verbal information on things we do
not presently perceive or infer ourselves, sabdah sabdyate
anenartha ity abhidhlyate jnapyate, 'Reliable statement:
by it an object is verbally expressed, thus it [the
object] is pointed out, made known (NBh 24.4) *.
1.3 Perception
Every form of cognition starts with the direct perception
of an object by the senses. In so far as perception con-
veys true knowledge, it is to be regarded as a means of
valid cognition.
indriyarthasannikarsotpannaiTi jnanam avyapadesyam
avyabhicari vyavasayatmakam pratyaksam,
'Knowledge which arises from the contact of the senses
with an object, is perception [as a means of valid cog-
nition, in so far as that knowledge is] not [yet] express-
ible [in language], not erroneous and consisting in [the]
ascertainment [of what that object really is] (NS 1.1.4)1.
It is clear from the Sanskrit-sentence that sensory per-
ception, pratyaksa, is primarily defined as knowledge,
j_nana_, that issues from the senses and their objects, and
secondarily (but no less important) as not expressible in
language, avyapadesya, etc. The contact of the senses with
an obj ect,indriyartha-sannikarsa, determines knowledge,
such knowledge determines sensory perception in general,
while 'not expressible in language', 'not erroneous' and
'consisting in ascertainment* determine the aforementioned
perception as a means of valid cognition, pramana.
In Gupta 1962, p. 26-27 it is suggested that this
definition originally could have run:
indriyarthasannikarsotpannaiti jnanam pratyaksam,
while the other three qualifications were added later. It
cannot be established with absolute certainty that this
has really happened, but Gupta makes out a good case by
referring to the older perception definition in Vaisesika
Sutra III.1.18,
atmendriyarthasannikarsad yan nispadyate tad anyat,
"Was aus dem Kontakt von Seele, Sinnesorgan und Gegenstand
entsteht, das ist etwas anderes (als die Schlussfolge-
rung)" (Gupta 1962, p. 27).
It seems to me that NS 1.1.4 also defines a means of
cognition in general as not erroneous and consisting in
ascertainment. The first qualification avyapadesya, 'not
expressible in language' can only refer to pratyaksa as
8
pramana, since the knowledge produced by the other means
of valid cognition is of necessity expressible and ex-
pressed in language, viz the knowledge produced by in-
ference, analogy and reliable statement. However, all
four pramanas have to be non-erroneous and consisting in
ascertainment, for if they were not so, they would cease
to be means of valid cognition.
In the NBh the three specific qualifications of per-
ception as pramana are explained one by one. About the
condition that perception should be a mode of knowing that
is not yet expressible in language Paksilasvamin writes;
yavadartham vai namadheyasabdas,tair arthasampratyayah,
arthasampratyayac ca vyavaharah. tatredam indriyartha-
sannikarsad utpannam arthajnanam rupam iti va rasah ity
evam va bhavati, ruparasasabdas ca visayanamadheyam. tena
vyapadisyate jnanam rupam iti janite rasa iti janlte
namadheyasabdena vyapadisyamanam sat sabdam prasajyate,
ata aha avyapadesyam iti
(NBh, p. 28.1-29.1), 'Indeed, there are [so many] words
[being] names [signifying objects] as there are [real]
objects. By means of these [denotative words alone] [the
perceiver gets] the [firm] cognition of objects and from
the [firm] cognition of objects [proceeds on the part of
the perceiver intentional] activity [with regard to these
objects; therefore direct perception of objects is always
determined by the denotative words for the objects]. As
regards this [contention we say,] the object-knowledge
that has arisen from the contact of a sense with an object
thus becomes [afterwards for the perceiver the internal
notion] "form" or "taste", and the words for form and
taste are the [denotative] name of a [real, outer] thing.
By the [denotative name] [perceptual] knowledge is expres-
sed, [ie the perceiver] knows for himself "[this is a]
form" or "[this is a] taste". While [this perceptual know-
ledge] is being expressed through a word which is a name
[denoting an outer object], it [ie perceptual knowledge]
[could wrongly] be considered verbal [knowledge, ie know-
ledge determined fully by words and not by the outer ob-
jects at all]. Therefore [in order to avoid such a mis-
taken view of perceptual knowledge] he [Aksapada] says
"not [yet] expressible in language".
At the beginning of this passage it is suggested that
knowledge is, and can only be, received through language,
ie that there is no form of direct perception in which
language is not involved. But the Bhasya refutes this view
by saying that the impressions derived from the senses
receive their names like "form" and "taste" afterwards in
the mind of the perceiver. Those sensory impressions are
the objects that are verbally designated, but they have
no direct connexion with the words that denote them. In
order to distinguish perceptual knowledge from all the
other modes of knowing in which language jjs necessarily
involved, it is said that perceptual knowledge is not ex-
pressible in language, ie it is an immediate and purely
non-verbal form of cognition of outer objects.
In order to be pramana this purely non-verbal
knowledge should not be erroneous:
grisme marlcayo bhaumenosmana samsrstah spandamana
durasthasya caksusa sannikrsyante, tatrendriyarthasanni-
karsad udakam iti jnanam utpadyate, tac ca pratyaksaiti
prasajyate ity ata aha avyabhicarlti.
'In the hot season, rays of sunlight which are connected
with the heat of the soil and are vibrating [because of
the heat], come in contact with the eye of a [person, an
observer] standing at a great distance. [And] there [in
the observer] on account of the contact of the senses
with the object, arises the judgement "water". Now that
can also [wrongly] be considered perception [as pramana],
therefore [in order to avoid this wrong idea] he [Aksa-
pada] says "non-erroneous" (NBh 30.1-3)'. Perceptual
judgements that contain errors are obviously ineffective,
as they give no valid information at all.
The last special characteristic of perception is its
consisting in ascertainment:
durac caksusa hy ayam artham pasyan navadharayati dhuma
iti va renur iti va» tad etad indriyarthasannikarsotpannam
anavadharanajnSnam pratyaksam prasajyata ity ata aha
vyavasayatmakam iti.
10
1 [If ] this [observer] at a great distance is watching with
his eye a thing and does not determine whether it be smoke
or dust, [then] such indeterminate knowledge springing from
contact of the senses with an object can [wrongly] be con-
sidered perception [as pramana]; therefore [in order to
avoid this error] he [Aksapada] says "consisting in ascer-
tainment" (NBh 30.5-7) '. As long as a sensory impression
remains indecisive, it cannot function as an instrument of
valid knowledge and such indeterminate knowledge would be
quite unsuitable for procuring what is desirable and
avoiding the undesirable (cf NBh 1.1-2 quoted above).
1.4 Inference
We have already seen Paksilasvamin's description of in-
ference as 'the subsequent cognition of an object with the
help of a property [of that object] which is [already]
known through a means of valid cognition (NBh 24.1 )'. Now,
first we must look at the cryptic NS-definition of in-
ference, which as yet seems to elude explanation. It runs:
atha tatpurvakam trividham anumanam purvavac chesavat
samanyatodrstam ca,
'And preceded by this [ ie by sensory perception] there is
three-fold inference, [namely inference] possessing [ ie
concerning] what was previous, possessing [ie concerning]
the rest, and [inference based on that which is] seen from
similarity (NS 1.1.5)'. This translation is only tenta-
tive. The first part of the definition does not present
too many difficulties^2. It is both conceivable and ra-
tional that inference takes place only on the basis of
correct knowledge acquired through the senses, so that
inference is 'preceded by that1, tatpurvakam. Paksila-
svamin *s commentary on the initial part of this definition
is easy to follow.
tatpurvakam ity anena lingalinginoh sambandhadarsanaih
lingadarsanam cabhisambadhyate,
'Preceded by that, this refers to seeing the connexion be-
tween a property and the property-bearer and to seeing the
property [connected with the property-bearer) (NBh 33.3) '.
11
Thereupon Paksilasvamin depicts the process of inference
itself:
lingalinginoh sambaddhayor darsanena lingasmrtir
abhisambadhyate. smrtya linqadarsanena capratyakso1rtho 'numlyate,
'The recollection of the property is related to [the fact
that one] has seen that the property and the property-
bearer are connected [with each other]. By bringing [that
connexion] back to memory and seeing the property, an in-
visible object, [ie the presently invisible property-
bearer] is inferred (NBh 33.4-5)'. If we would substitute,
for example, fire for property-bearer and smoke for
property, this assertion becomes clear. At a certain time
we saw for ourselves that smoke and fire are connected; we
have often observed fire producing smoke . If we remember
this, and after a while we see some smoke but no fire, we
may infer the presence there of fire on the grounds of
smoke. In short, we see thing A and thing B connected, the
connexion being that of property-bearer and property. When
at a certain time we see thing B alone, we infer thing A
to be present as well. In the case of smoke and fire,
smoke is considered to be the property of fire. One might
question the admissibility of these substitutions, but
consider the following use made by Paksilasvamin of the
terms property, linga and property-bearer, lingin.
apratyakse lingalinginoh sambandhe kenacid arthena
lingasya samanyad apratyakso lingi gamyate, yatheccha-
dibhir atma, icchadayo gunah gunas ca dravyasamsthanah,
tad yad esaiti sthanam sa atmeti,
'When the connexion between property and property-bearer
is not directly perceivable, a not-perceivable property-
bearer is cognized in so far as the [same] property is
common to some [other visible] object, as for instance:
the soul [is connected] [with perceivable properties such
as] desire etc. Desire etc. are [perceivable] attributes
and [all] attributes have as their abode a substance. The
locus of these [attributes such as desire etc.] is the
soul [hence it is proved that the soul exists] (NBh 36.2-
4) f . The property-bearer, lingin, in the example is the
12
soul, while the properties, lingani, are the soul's
attributes, gunah, such as desire etc. . In a syllogism
one would have to predicate the possession of a soul of
some subject; eg this man possesses a soul on the grounds
that he shows signs of desire etc, which must be regarded
as attributes necessarily inherent in a substance. In the
syllogism (as will be made clear later), the property-
bearer and its properties are themselves considered pro-
perties of a subject, sadhya, of which some property is
inferred on the grounds of the subject's possessing an-
other property connected with the first property. So far,
inference as described in the NS and the NBh does not give
great difficulties of interpretation. The interpretation
of NS 1.1.5 becomes problematic with the term 'threefold1.15
This qualification is not explained in the Bhasya, so we
may assume that Paksilasvamin takes it to refer to the
three qualifications enumerated in the sequel of this
aphorism.
As for the terms 'possessing what was previous1,
purvavat, 'possessing the rest',sesavat, and 'seen from
similarity1, samanyatodrsta, Paksilasvamin gives two
divergent explanations. A remarkable fact indeed, and in-
dicative of the confusion that existed about these three
terms already in Paksilasvamin's time. The first set of
explanations runs as follows:
purvavad iti, yatra karanena karyam anumlyatef
'Possessing what was previous, when an effect is inferred
from a cause (NBh 34.1)'.
sesavat tat, yatra karyena karanam anumlyate,
'Possessing the rest is that [inference] in which a cause
is inferred from an effect (NBh 34.2)*. Samanyatoidrstfr,
'seen from similarity1, is explained thus:
samanyatodrstarn, vrajyapurvakam anyatra drstasyanyatra
darsanam iti, tatha cadityasya tasmad asty apratyaksapy
adityasya vrajyeti,
'Seen from similarity: seeing at one place an [object that
was previously] seen at another place [while this seeing
it now] is preceded by the movement [of the object] and
thus, [seeing that] the sun [has changed position, one
13
infers] that on those grounds also the sun possesses mo-
tion [albeit] not immediately perceived (NBh 34-3.5)'. If
objects are seen to always be in motion when they change
places, it is possible to infer the motion of the sun be-
cause it changes its position in the sky during the day.1 Similarity', in all cases, is the observed change of po-
sition, and from this fact motion is inferred of things
that are not actually seen to be moving.
In the next set of explanations Paksilasvamin inter-
prets purvavad quite differently.
athava purvavad iti yatra yathapurvam pratyaksabhutayor
anyataradarsanenanyatarasyapratyaksasyanumanam yatha
dhumenagnir itiy
(NBh, p. 34.6-35.1), 'Or purvavad [means: that case] in
which -- by seeing one of two [objects] of direct percep-
tion in the way (yatha) [they have] previously (purvam)
[been seen] — the other [thing that is now] not perceiv-
able is inferred [to be present], as [the presence of]
fire [that is now not seen is inferred] by [seeing]
smoke*.
This explanation runs opposite to the first, which
said purvavad meant inferring an effect from a cause. Ac-
cording to this second explanation, the "previous", purva,
refers to the fact that previously two things were seen
together, while the presence of the one thing that is now
not directly perceived is inferred from the presence of
the other connected thing that ijs perceived.
As regards sesavad, too, Paksilasvamin gives an alter-
native explanation.
sesavan nama parisesah sa ca prasaktapratisedhe 'nyatra-
prasangac chisyamane sampratyayah,
'Possessing the rest means the remainder, and that is the
firm conviction as regards [the only] remaining [logical
possibility] in so far as one other [of several possibili-
ties] is not absurd when [all] absurd [possibilities] have
been negated (NBh 35.1-2)'. In this explanation sesavad is
interpreted as "having the remainder" (sesa) in the con-
text of NS 1.1.5. Moreover, this alternative interpreta-
tion deviates very much from the previous one, according
14
to which sesavad refers to the inference of a cause from
an effect.
The second explanation of samanyatodrsta is given in
the passage already quoted in connexion with the terms
'property1 and 'property-bearer1 in NBh 36.2-4, There it
is maintained that a property-bearer which is not directly
perveived can be inferred to exist on the grounds that it
has properties similar to those of a perceivable property-
bearer. The "invisible" property-bearer is "seen" (drsta)
from its similarity to (samanyatah) "visible" property-
bearers. This explanation differs from the previous one of
samanyatodrsta; for in the first, in NBh 34.3-5, the
property-bearer was not invisible but a property of it,
motion, was invisible. In the second explanation the pro-
perties are visible but the property-bearer is not
visible.
In the concluding paragraph on inference, Paksila-
svamin remarks:
sadvisayam ca pratyaksam sadasadvisayam canumanam,
'The realm of perception is that which is [now manifestly]
present; the realm of inference is that which is [now]
present as well as that which is not present [now] (NBh
37.1 ) f .
kasmat ? traikalyagrahanat, trikalayukta artha anumanena
grhyante bhavisyatlty anumlyate bhavatlti cabhud iti ca,
asac ca khalv atltam anagatam ceti,
'Why ?, Because [inference] grasps the [totality of the]
three times. Objects connected with the three times [past,
present and future] are grasped by inference, [because] it
is inferred "[this] will be", "[this] is [now]" and
"[this] was", and indeed [a thing of] the past and [of]
the future are not present [now] (NBh 37.1-3)'. This seems
to summarize the discussion on the three types of in-
ference by suggesting that the threefoldness refers to the
fact that we can draw inferences about things in the past,
present and future16.
15
5 Comparison based on analogy
The third means of valid cognition is comparison based on
analogy, upamana. It is defined thus:
prasiddhasadharmyat sadhyasadhanam upamanam
(NS 1.1.6), Comparison [based on analogy] is the demon-
stration of a probandum [ie a thing that is to be demon-
strated or indicated] on the grounds that [the probandum]
possesses [in some degree] the same properties as a [thing
or property-bearer that is] well known ~[to the hearer to
whom the analogy is pointed out]1. Paksilasvamin has al-
ready briefly explained comparison in NBh 24.2 cited
above, 'Comparison [based on analogy] is knowledge of the
sameness of form: "a bos Gavaeus is like a cow1. On the
NS-definition itself he comments:
prajnatena samanyat prajnapanlyasya prajnapanam upamanam
iti. yatha gaur evam gavaya itif
'Pointing out of what is to be pointed out in so far as
the [thing to be pointed out] is similar to a known thing,
is [called] comparison [as for instance when we say]: "a
bos Gavaeus is like a cow" (NBh 37.6-7)'. And moreover:
yatha gaur evam gavaya ity upamane prayukte gava samana-
dharmam artham indriyarthasannikarsad upalabhamano ^sya
gavayasabdah sanjneti sanjnasanjnisambandham pratipadyata
iti,
'After the comparison "a bos Gavaeus is like a cow" has
been employed [by a speaker], a person who cognizes from
the contact of the senses with the object [ie by pra-
tyaksa] a thing whose properties are similar to those of
a cow, knows the connexion between the designation and the
designated, namely: the word bos Gavaeus is the designa-
tion of this [cowlike animal] (NBh 38.2-3)'• Comparison,
or rather, gaining knowledge of an unknown thing through
testimony and the perceivable resemblance of the unknown
thing to a known thing does not constitute what at first
sight seems to be an inference. For by inference we come
to know a hitherto unknown property of a given subject
from seeing another intimately connected property. And
since we observed that the two properties were always
16
connected, we legitimately infer the first property to be
present in the subject. The two properties themselves are
known to us, the subject is known, but the relation of one
of the properties to the subject is not known; it is in-
ference that makes such a relation known. As for compari-
son, we know one particular object and its properties,
whereas there may be a multitude of objects that partly
resemble our known object. Another expert person has to
point out to us such an unknown object by saying it great-
ly resembles the known thing. This speaker must of neces-
sity know both objects. We know only one, but as soon as
we see an object closely corresponding to the speaker's
description (and our recollection of the known thing), we
know that we have discovered the unknown thing. Through
comparison we proceed from a known object to an unknown
object on the strength of our perception and reliable
testimony, whereas in inference we proceed from two known
objects to an unknown relation between one of these two
with a third known object. The basis of inference and com-
parison remains sensory perception, the fundamental means
of valid cognition.
1.6 Reliable Statement
The fourth and last means of valid cognition is a way of
knowing objects solely by verbal designation. Of course
the perceiver must perceive the statements which contain
this knowledge with his own organ of hearing, but the new
knowledge that is brought to him does not come into being
through his own sensory perception as in inference and
comparison. The latter could be regarded as a combination
of perception and a reliable statement, because the first
element in that cognitive process, the object to which the
unknown object is compared (by a reliable speaker) is
known, and needs to be known, through sensory perception.
In the case of a reliable statement, the object is fully
unknown to the hearer and the validity of the knowledge
rests entirely on the validity of the statement and hence
on the reliability of the speaker.
17
NS 1.1.7 defines reliable statement as a means of
valid cognition in this way: aptopadesah sabdah,
'[Reliable] statement is the instruction of an expert1.
The word apta has various meanings among which 'expert1,
'trustworthy1, 'reliable' and 'skilful'. It is intended in
this aphorism that the reliability of the statement or in-
struction depends on the trustworthiness of the speaker
and his expertise. Moreover, it is quite clear that not
simply any statement could be pramana, but only those that
contain true information in the form of valid instruction.
Paksilasvamin writes about this aphorism;
aptah khalu saksatkrtadharma yathadrstasyarthasya cikhya-
payisaya prayukta upadesta. saksatkaranam arthasyaptih.
taya pravartata ity aptah. rsaryamlecchanam samanaiti
laksanam. tatha ca sarvesam vyavaharah pravartanta iti,
'Now, an expert is an instructor who has directly per-
ceived the dharma [righteousness, the moral order in the
cosmos] and who is urged by a desire to describe a thing
as he has seen it. Expertise concerning a thing is the
direct perception [of it] and he who acts with that [ex-
pertise] is an expert. [Expertise can be] the common char-
acteristic of sages, honourable people [following the or-
thodox way of life according to the Vedas] and barbarians
[those that do not follow the orthodox way of life, ie the
foreigners]. And [it is a fact that] in this way [with the
help of expertise] the [daily] affairs of all [beings] are
carried out (NBh 40.1-3)'. A person who has directly per-
ceived the dharma, a saksatkrtadharma, seems to denote
Vedic rsis, for the term occurs already with Yaska, the
author of the Nirukta (cf Chemparathy 1983, p. 21, 78 note
20 which mentions some relevant literature on dharma and
Oberhammer 1974, p. 50, note 29 on saksatkrta-dharmata-).
In the broader context evidently intended in the NBh,
dharma seems simply to mean 'the true state of a thing or
fact1, 'the true nature of things'. The immediate ex-
perience of a thing or fact constitutes the expertise of
the expert. In a rather democratic spirit the commentator
acknowledges that all sorts of people, from Vedic seers to
barbaric foreigners, can be an expert in something. In ac-
18
complishing many of our everyday concerns, we need practi-
cal information in the form of reliable statements made by
those who know something thoroughly. And yet reliable
statement, sabda, does not denote only practical informa-
tion. The sutra says:
sa dvividho drstadrstarthatvat,
'it [reliable statement] is of two kinds according to
whether the object [it refers to] is [something] manifest
[to the senses] or not manifest [to the senses] (NS
1.1.8)'. Paksilasvamin comments:
yasyeha drsyate frthah so drstarthah. yasyamutra pratlyate
so 'drstarthah. evam rsilaukikavakyanam vibhaga iti,
'If the object of a [statement] is [something] seen here
[on earth], the [statement] refers to [something] manifest
[to the senses]. If the [object] of a [statement] is ex-
perienced in the other world, the [statement] refers to
[something] not manifest [to the senses]. In this way we
divide the assertions of sages and ordinary people (NBh
41.1-2)f. The commentator continues:
kimartham punar idam ucyate ? sa na manyeta drstartha
evaptopadesah pramanam arthasyavadharanad iti; adrstartho
'pipramanam arthasyanumanad itif
'For what purpose is this [division] enunciated ? One
should not think that the instruction of an expert is a
means of valid cognition only [if] [the instruction]
refers to something manifest because the object of the
statement can be ascertained directly [by the senses].
Also a [statement] referring to something not manifest [to
the senses] is a means of valid cognition because the ob-
ject [of the statement can] be inferred [to exist] (NBh
41.2-4)'. The first kind of reliable statement gives in-
formation about visible thingsf and these statements can
be validly made by both proficient sages and proficient
common persons. We are capable of verifying such state-
ments, as the objects they describe are always somehow
visible to our senses. Thus, it is said that the object of
this kind of statement can be ascertained directly. The
second kind gives information on invisible things, espe-
cially of course on religious topics such as the hereafter
19
and final liberation from suffering. It is not possible
for us to verify such statements directly by the senses,
because the objects these statements refer to elude our
perception. Hence it is maintained that the existence of
those invisible objects can only be proved through in-
ference, which probably means that we must infer the ex-
istence of those objects from the trustworthiness of the
ones who speak about them as is asserted later in NBh
225.1-2.17
Observing the four pramanas in the given order of
perception, inference, analogy and reliable statement, one
discovers an increasing remoteness on the part of the per-
ceiver from the object to be cognized. This would explain
the particular order in which they are treated in the NS.
This order implicitly evinces the increasing distance of
the perceiver from the object, besides demonstrating the
prime importance of direct perception in knowing reality.
Perception is the most direct and closest way of cognizing
an object. Inference is a correct but indirect form of
cognition in which perception plays some part so that it
is not as direct as perception. Comparison is still less
direct than inference, since it requires the reliable tes-
timony of another for its correct functioning, unlike in-
ference which can be performed by oneself. Reliable state-
ment functions entirely without our own perception, unlike
comparison which requires that we perceived the object
with which the unknown object is compared. We cannot per-
form the latter two acts of cognition without the outside
help of a reliable speaker, while we can correctly carry
out the former two acts of cognition by ourselves • Hence
we can say that the order of pramanas, beginning with
perception, suggests a decreasing directness, which means
a greater cognitive distance from the object to be cog-
nized. It is almost a truism that an accurate description
of an object in no way equals the direct sensory percep-
tion of the object; for an accurate description is what
reliable statement, sabda, really amounts to.
20
7 The Syllogism
It is through philosophical debate that the true nature of
a fact is ascertained:
vimrsya paksapratipaksabhyam arthavadharanam nirnayah,
Ascertainment is determining a thing [concerning which]
doubts have arisen [; determining is done in discussion]
by means of a [philosophical] position and an adverse po-
sition (NS 1.1.41)1. Ideally, a philosophical debate pro-
ceeds in the form of a discussion (vada) in which deduc-
tive arguments, syllogisms, are put forward;
pramanatarkasadhanopalambhah siddhantaviruddhah
pancavayavopapannah paksapratipaksaparigraho vadah,
'Discussion consists in keeping up [a thesis] by means of
a [philosophical] position and an adverse position [; in
this keeping up] there is the establishing [ ie the affir-
mation of something] and censure [ie the negation of some-
thing] through the means of valid cognition and reasoning
[meant to know the truth about a knowable thing; and fur-
ther, this keeping up a thesis] is not contrary [ ie ought
not to be contrary] to established tenets and [it] is ac-
companied by the five members [of the syllogism] (NS
1.2.1 )*. In this definition of a genuine philosophical in-
quiry meant to find the truth about a thing, the instru-
ments for getting valid knowledge, the four means of valid
cognition, as well as the syllogism are explicitly men-
tioned. Even though, ultimately, the means of valid cogni-
tion produce all true knowledge, the syllogism is in fact
-- according to this definition -- the moving force of
dialectically putting forward a doctrinal point of dis-
pute, for it is said that this putting forward, keeping
up, parigraha, is accompanied by, upapanna, the five mem-
bers of the deductive syllogism. Deduction means validly
inferring a particular fact from given general facts which
function as major and minor premises. It is regarded in
the Nyaya as the supreme kind of reasoning, paramo nyayah
(NBh 12.1). Paksilasvamin writes:
kah punar ayam nyayah ? pramanair arthaparlksanaitt
nyayah. pratyaksagamasritam anumanaft sanvlksa.
21
'But what is this nyaya (sound reasoning)? Nyaya (sound
reasoning) is investigating an object with the means of
valid cognition. Such investigation is [actually] in-
ference based on perception and tradition [which when
valid is reliable statement] (NBh 8.3-4)'. These words
signify that the correct syllogism is an explicit
inference.
A complete syllogism consists of five members:
pratij nahetudaharanopanayanigamanany avayavah,1 The members [of the syllogism] are: thesis, reason,
example, application and conclusion (NS 1.1.32) " . I will
now give the NS' s definitions of these terms along with
some of Paksilasvamin1s clarifications.
sadhyanirdesah pratijna,
'The thesis means indicating what is to be demonstrated
(NS 1.1.33)1. The thesis should specify the subject and
the property of which it must be proved that it occurs in
the subject: in other words, the thesis is a proposition
which specifies the subject and the probandum.
prajnapanlyena dharmena dharmino visistasya
parigrahavacanam pratiina,
'The thesis is keeping up [as a hypothesis] that a par-
ticular property-bearer is possessed of a property that is
yet to be shown [to occur in that property-bearer] (NBh
81.7)'. And elsewhere, Paksilasvamin says:
sadhyasya dharmasya dharmina sambandhopadanaiTi
pratij narthah y
'The purport of the thesis is accepting that there is a
connexion of a property-bearer with a property that is to
be proved [to occur in the property-bearer] (NBh 91.3)'.
The example of a thesis given by Paksilasvamin is:
anityah sabda iti,
'Sound is non-eternal (NBh 81 .*8 ) ' . This necessarily means
all sounds, but unlike Aristotelian logic, NS- and NBh-
logic do not use an explicit quantifier. Here, sound is
the property-bearer, while non-eternity is the property of
which we wish to prove that it occurs in the property-
bearer, sound.
As such, this statement is not yet a piece of solid
22
knowledge? for being that it requires a valid logical
reason, hetu. The next sutra defines the reason:
udaharanasadharmyat sadhyasadhanam hetuh,
'That [property, probans] which demonstrates the probandum
is the logical reason in so far as the same property [the
probans] occurs in the example (NS 1,1.34)'. It is neces-
sarily implied here that the probans (hetu) occurs in the
subject of which the other property, the probandum, must
be proved to occur in it. If the probans would not be a
property of the subject, the whole reasoning of the syl-
logism becomes nonsensical. The example in which the
probans must occur too, is a commonly acknowledged fact,
drstanta; in the drstanta brought forward as exemplifica-
tion in the syllogism, the probans and probandum are seen
together as related or connected (sambaddha) properties .
The commentary states:
udaharanasamanyat sadhyasa dharmasya sadhanam prajnapanam
hetuh, sadhye pratisandhaya dharmam udaharane ca pratisan-
dhaya tasya sadhanatavacanam hetuh. utpattidharmakatvad
iti. utpattidharmakam anityam drstam iti,
'The reason is that which demonstrates [or] shows the pro-
perty that is to be proved through similarity with the
example; the reason is the verbal expression of the fact
that the [particular property now occurring in the sub-
ject] is the probans, after it is remembered that the pro-
perty [the probans] occurs in the subject and after it is
remembered that it [also] occurred in the example. [As in
the illustrative syllogism: sound is non-eternal] because
it possesses the property of production [ie of being pro-
duced]. It is seen [and hence well-known] that [every-
thing] which has the property of production is non-eternal
(NBh 82.2-4)', Paksilasvamin clearly states that the pro-
bans, the hetu, is present in the subject. In the illus-
trative syllogism, 'the property of production1 serves as
probans; 'being non-eternal1 is the probandum. Whatever is
produced is acknowledged by all to be ephemeral, non-
eternal. The only logical connexion between probans and
probandum, according to Paksilasvamin, is the fact that
both are always seen (drsta) together in the commonly ac-
23
knowledged fact (drstanta) serving as example (udaharana).
In order to be valid, the reason must not occur in
examples in which the contrary of the probandum is found.
Thus, for example, the property of production which is
probans, must not be present in things that are eternal,
which is the contrary of being non-eternal. Hence the NS
declares: tatha vaidharmyat, 'And in so far as that
property [ie the probans] does not occur [in a contrary
example] (NS 1 • 1•35) *• We have to supply the beginning of
NS 1.1.34 to give the proper sense to this sutra: 'That
[property] which demonstrates the probandum is the reason
[also] in so far as this property does not occur [in a
contrary example]'; this would have been the complete
formulation.
Paksilasvamin explains it at some greater length:
udaharanavaidharmyac ca sadhyasadhanaffl hetuh,
'The reason is that which demonstrates the probandum also
in so far as this property [the probans] is absent from
[contrary] examples (NBh 83.2)'. For if the probans would
also occur with a property contrary to the probandum, the
probans would never conclusively prove the probandum and
consequently the syllogism would sometimes fail.
Paksilasvamin continues:*
katham anityah sabdah utpattidharmakatvat.
anutpattidharmakam nityam yathatmadi dravyam iti,
'How [as in our illustrative syllogism:] is sound non-
eternal?; because it possesses the property of production.
That which is eternal does not possess the property of
production such as [all the unborn, eternal] substances
beginning with the soul (NBh 83.2-3)'. Paksilasvamin im-
plies here that being a product and being impermanent al-
ways go together, and that being unborn and being eternal
always go together, whilst being a product and being per-
manent or being unborn and being non-eternal never go
together.
The validity of the syllogism is based on the example
in which a commonly known fact is put forward in support
of the logical reason. The definition of the example is as
follows.
24
sadhyasadharmyat taddharmabhavi drstanta udaharanam,fA generally-accepted fact which possesses the property
[to be demonstrated, the probandum] of the [subject] is a
[valid] example [in the syllogism] in so far as the same
property [the probans being a known property] of the sub-
ject occurs [in the generally-accepted fact] (NS 1.1.36)1.
In this aphorism it is emphasized that the probans (hetu)
must be found in the subject, as well as in the positive
example where it occurs with the probandum. The positive
example is a drstanta which is positive because of the
following two facts: a. in it occurs the probandum
(taddharmabhavin); b. in so far as it also possesses the
probans, a known property of the subject (sadhyasa-
dharmyat ), this drstanta can function as a valid, true,
and positive example in reasoning. The ablative sadhya-
sadharmyat expresses the condition under which the
drstanta is positive, namely: the drstanta which possesses
the probandum should also possess the probans. The example
is not simply an analogous instance cited to make the
thesis more plausible, but the mentioning of a fact that
is acknowledged to be true by all. One is tempted to say
that what is implied here is: a drstanta is a fact con-
sisting in an indefinite number of cases where the probans
and the prodandum have been seen together, not a fact con-
sisting in one single case in which the probans and the
probandum have been seen together. But this interpretation
of drstanta is not explicitly given in the NS, even though
it is a very likely one. It seems that the syllogism as
outlined in the NS itself already offers the framework of
a sound deductive proof.
The phrase sadhyasadharmyat Paksilasvamin glosses as
follows:
sad-hyena sadharmyam samanadharmata,1Sadharmya with the subject means being [endowed] with the
same property [as the subject] (NBh 83.5)'. The fsame pro-
perty1 refers to the known property of the subject; this
known property acts as logical reason because it is seen
together with the property to be proved. In the drstanta
we notice that it is always the case that whenever there
25
is the probans there is the probandum.
The phrase taddharma is glossed*
tasya dharmas taddharmah. tasya sadhyasya,1Taddharma [means] the property of that; of that [means]
of the subject (NBh 84.1 ) f • The property of the subject
here refers to the property that is to be proved, of which
it is said that it occurs in the drstanta. The definition
in the sutra itself is constructed as follows: the
generally-accepted fact has the probandum, and because it
also has the probans, it serves as a valid example in the
syllogism. To substantiate the importance of the drstanta
in syllogistic, deductive reasoning, I will quote the de-
finition of drstanta given in the NS with Paksilasvamin's
explanation. The sutra defines,
laukikapariksakanam yasminn arthe buddhisamyam sa
drstantah,
'A generally-accepted fact is a thing of which ordinary
people and [learned] investigators have the same cognition
(NS 1.1.25)1. Thus a drstanta is a thing, situation or
fact that is accepted to be really the case by non-experts
and experts alike. The term drstanta by itself already
evokes this idea of universal acceptance, for it literally
means ascertainment, definite settlement (anta) seen, di-
rectly perceived, acknowledged (drsta) [by all]. For
example, everyone, non-experts and experts, knows that
wherever there is smoke there is fire, or that everything
created must some day perish, is non-eternal.'
Paksilasvamin says somewhere about drstanta:
atha drstantah pratyaksavisayo 'rthah. yatra laukikapari-
ksakanam darsanam na vyahanyate,8Now drstanta is a thing which is an object of direct per-'
ception concerning which the vision of both ordinary
people and [learned] investigators is not confused (NBh
10.3-4)'. And a little further he describes the importance
of the drstanta:
tadasrayav anumanagamau. tasmin sati syatam anumana-
gamav asati ca na syatam,
"inference and tradition [consisting of reliable state-
ments] are both based on a [drstanta].
26
When this [drstanta] is present, inference and [reliable]
tradition are possible [ie can be regarded as pramanas],
when the [drstanta] is absent, [inference and reliable
tradition] are impossible (NBh 10.4-5)'• These two pas-
sages explicitly state that a drstanta should be used in
drawing a correct inference; this means that a syllogism
(in which drstantas are formulated) could simply be con-
sidered a verbalized inference. The idea that reliable
statements (sabda, agama) too, should be based on a
drstanta suggests that the ideal reliable statement is not
different from syllogistic reasoning.
In the commentary on NS 1.1.25 Paksilasvamin
delineates the function of drstanta in dialectical discus-
sions and the syllogism;
drstantavirodhena hi pratipaksah pratiseddhavya
bhavantiti drstantasamadhina ca svapaksah sthapanlya
bhavantlti; avayavesu codaharanaya kalpata itiy
'For, the opposite positions must be contradicted through
[their] disagreement with a generally-accepted fact, and
one's own positions must be established through [their]
agreement with a generally-accepted fact; and [moreover]
in the syllogism [the generally-accepted fact] adequately
serves as the example (NBh 73.2-3)'.
Returning to the discussion on the example, we find
that Paksilasvamin shows its use in the illustrative
syllogism: sound is non-eternal because it has the
property of production.
sabdo fpy utpattidharmakatvad anityah sthalyadivad iti.
udahriyate dharmayoh sadhyasadhanabhava ity udaharanam,
'Also sound, because it possesses the property of produc-
tion, is non-eternal like [all produced] things [such as]
vessels etc. By this [well-known fact] it is exemplified
that the two properties function as probandum and probans;
in this manner the example [works] (NBh 85.1-2)'. The
well-known fact here, is that all visible and tangible ob-
jects like cooking vessels etc. are seen to be made by
someone, and are seen to be subject to decay. It is im-
plied although not clearly stated that the two properties
are invariably concomitant .
27
To the phrase in NS 1.1.35 'And in so far as that property
[the probans] does not occur [in a contrary example]1 cor-
responds a definition of the contrary example.
tadviparyayad va viparltam (NS 1.1.37), f0r [the example
is] contrary in so far as the [generally-accepted fact
constituting the example] is contrary [ie a generally-
accepted fact which does not possess the probandum is a
valid contrary example in so far as the probans does not
occur in it either]1. It has been said in NS 1.1.35 that
the probans should never occur with properties contrary to
the probandum, which is the same as saying that the pro-
bans should not occur without the probandum. The contrary
example shows that when the probandum is absent, the pro-
bans is likewise absent. Paksilasvamin illustrates this
through the familiar syllogism on sound:
anityah sabda utpattidharmakatvat. anutpattidharmakam
nityam atmadi. so 'yam atmadir drstantah sadhyavaidharmyad
anutpattidharmakatvad ataddharmabhavl; yo 'sau sadhyasya
dharmo 'nityatvam sa tasmin na bhavatlti,
'Sound is non-eternal because it possesses the property of
production. [Every substance that] is eternal [like] the
soul etc. does not possess the property of production.
This generally-accepted fact, starting with [unborn, eter-
nal substances like] the soul, while not possessing the
property [the probandum] of the [subject, namely non-
eternity] is [a valid logical example] in so far as in it
[ie in the drstanta] does not occur the property [the
probans] of the subject, [that is to say] the property of
production; ie this property [the probandum ie] non-
eternity of the subject is not present in this [drstanta]
(NBh 86.1-3)1.21
The fourth step in the syllogism is called upanaya,
'application1•
udaharanapeksas tathety upasamharo na tatheti
va sadhyasyopanayah,
'The application of the subject is [its] summing up based
on the [positive and contrary] examples: either [some-
thing] is thus [the case], or it is not (NS 1.1.38)1. In
the application, we must ascertain whether the property,
28
being present in the subject, is really the same property
(as probans) we saw in the positive examples to be con-
nected with the probandumf and in the contrary examples to
be never connected with the contrary of the probandum,
which implies that when the probandum is absent, the pro-
bans is absent. Application means that the general rule of
the drstanta and the particular fact of the hetu is ap-
plied to, literally brought near (upa and verb, root nlE),
the subject, sadhya; the latter is compared to the posi-
tive examples and found to possess the same property (pro-
bans ) or compared to the contrary examples found to be to-
tally different from them. In the Bhasya it is shown how
the application is used,1••.sthalyadi dravyam utpattidharmakam anityam drstam
tatha sabda utpattidharmaka-iti sadhyasya
sabdasyotpattidharmakatvam upasamhriyate ffIt is seen that a substance [like] a vessel etc. which
is non-eternal has the property of production, [and]
thus [in the same way], sound has the property of produc-
tion; in this manner, the fact that sound, the subject,
possesses the property of production is summed up (NBh
87.3-4)'. This is the application based on a positive
example; the application according to a contrary example
runs as follows:
atmadi dravyam anutpattidharmakam nityam drstam na tatha
sabda iti: anutpattidharmakatvasyopasamharapratisedhenot-
pattidharmakatvam upasamhriyate,
'It is seen that a substance [like] the soul etc. which is
eternal does not have the property of production, [and]
sound is not like that; in this way, the fact that [sound]
has the property of production is summed up by means of a
negation, being the summing up of the fact that
[substances contrary to sound] possess the property of
non-production [ie are never created] (NBh 87.5-6)'.
The last step in the syllogism is the conclusion:
hetvapadesat pratijnayah punarvacanam nigamanam,
'After [again] pointing out the reason, the repetition
of the thesis [becomes] the conclusion [of the syllogism]
(NS 1.1.39)'. In the application it was decisively estab-
29
lished that the subject resembles the positive examples in
so far as the probans occurs in all; now it remains to
come to the logical conclusion of the syllogism. By point-
ing out the reason — the probans which was seen to be
connected with the probandum — and by stating that a cer-
tain subject is endowed with that probandum, the unfounded
thesis is made into a proved assertion, a true proposi-
tion. Paksilasvamin shows the workings of the conclusion
in his syllogism on sound:
tasmad utpattidharmakatvad anityah sabda iti nigamanamy
'Therefore [ie on the strength of the application and
the drstanta], because it possesses the property of
production, sound is non-eternal, thus [runs] the conclu-
sion (NBh 88.3-4)'.
The complete syllogism on the perishableness of sound
has the following form:
thesis : sound is non-eternal,
reason : because it is produced,
positive example : every substance that is
produced is non-eternal,
counter example : every substance that is not
produced is eternal
application : it is known that sound is
produced, like all substances
mentioned in the positive
example, while it is also
known that every eternal sub-
stance mentioned in the con-
trary example is not produced
conclusion : therefore sound is non-
eternal, because it is
produced .
The deductive reasoning as set forth in this syl-
logism was not done in quite the same order in the mind of
the speaker of the syllogism when he first drew the in-
ference. He must have started from the generally-accepted
fact, ie an object of direct perception (pratyaksa),
remembered this perceived fact and when a certain thing
was not perceptible, he hit on the reason and consequently
30
inferred the imperceptible thing to be present. To give an
an example: it is common knowledge that smoke is always
produced by fire (this piece of knowledge is drstanta);
once a person who knows this sees smoke somewhere without
fire, he infers (hetu) the necessary presence of fire on
the strength of the remembered drstanta. When he expresses
his judgement in this syllogism, however, that same person
first states as a thesis the fact he inferred, for he will
wish to start by making clear what he is going to prove;
only then will he give the reason and the commonly-known
fact from which the conclusion, that was initially offered
as a thesis, is drawn. Thus the order of steps in the
Nyaya-syllogism is not a natural but an effective rhetori-
cal one.
1.8 Trustworthiness of the speaker
It was shown above that reliable statements (sabda) are
of two kinds. Firstly, statements about visible things,
statements that might ideally take the form of correct
syllogisms; and secondly, statements about invisible ob-
jects, statements that are incapable of direct verifica-
tion. Especially as regards the latter, the truth of the
statement is wholly dependent on the trustworthiness of
the speaker, whereas the truth of the former kind of
statement is verifiable by direct perception and by test-
ing the logical validity if the statement is a syllogism.
True statements on invisible things ultimately refer
to statements concerning religious beliefs and values that
go beyond the ken of the empirical world. These values,
like final emancipation and the liberation from suffering,
are propagated by teachers of religion who have to be re-
liable if their doctrines are to be taken seriously. For
the Nyaya-philosopher, the supreme religious values are
those that are laid down by the ancient seers ( rsi ) in
the sacred texts of the Vedas. Hence, the follower of the
Nyaya (in accordance with his own epistemology) will
regard these holy texts as the trustworthy utterances
(sabda) of reliable expert speakers (apta). The trustwor-
31
thiness of the Vedas rests upon the trustworthiness of the
ancient seers, while the trustworthiness of the latter
must be inferred from their moral excellence. Religious
authority can only come from ethical superiority and the
usefulness of the religous precepts. In general, religious
instruction will call attention to the supreme good of
complete freedom from suffering, and reveal the way to
that exalted goal. In NS 2.1.69 we find only one rationale
of religious authority: the salutariness of the religious
doctrine.
mantrayurvedapramanyavac ca tatpramanyam aptapramanyat,
'And the fact that the [Vedas] are a means of valid cogni-
tion [ie have authority] like the fact that incantations
and medical science are a means of valid cognition [ie
have effective authority, is derived] from the fact that
experts are a means of valid cognition (NS 2.1.69)*. The
authority of the Vedas and the authority of the medical
science and certain magical charms derive from the autho-
rity of their authors. Medical science and charms refer of
course to reliable statements about something manifest to
the senses (drstartha). The alleged effectiveness of a
charm or medical prescription can be tested in daily life,
its utility is perceptible to the senses. The Vedas are
reliable statements about what is not manifest to the sen-
ses (adrstartha). Both kinds have their proper utility;
charms and medical prescriptions are meant to ward off
temporary suffering, ie illnesses, while the Vedas are
meant to remove suffering altogether, in that they des-
cribe the road to final emancipation, the highest good
(nihsreyasa). The usefulness of these texts (Vedic as well
as medical) derives from the authoritativeness of the ex-
perts who compiled them. It is possible to interpret this
sutra, as Paksilasvamin did: the expert authors of* the
medical texts known as Ayurveda are also the authors of
the Vedas, and since the medical texts are useful in des-
cribing effective cures for diseases, the Vedic texts are
useful too, especially as regards final liberation from
suffering.
ya evapta vedarthanafo drastarah pravaktaras ca ta
32
evayurvedaprabhrtinam ity ayurvedapramanyavad
vedapramanyam anumatavyam itiffThe same experts, (that is) those who see the objects
[described] in the Vedas and who relate [them to others],
are those ones [that see and communicate] the medical
texts etc; thus the authority of the Vedas, [which is]
like the authority of the medical texts, is inferred [from
the authority of the reliable experts] (NBh 225.1-2)f.
In what does the reliability and authority of these
experts consist? This question is not put forward in the
NS; Paksilasvamin brings it forward and proposes to answer
it. According to him, authority (pramanya) comprises:
saksatkrtadharmata bhutadaya yathabhutarthacikhyapayiseti,
'The fact of having directly perceived the dharma [the
true state of things], compassion for living beings, and
the desire to communicate the thing as it really is (NBh
223.5)'. The reliable authoritative speaker must have per-
ceived a particular thing for himself, he must know it
through his own direct perception. Further, he must have
compassion on others2**; and thirdly, through compassion he
wishes to inform others of his knowledge exactly as that
knowledge is. Only such experienced experts are assumed to
give good advice to people.
aptah khalu saksatkrtadharmana idam hatavyam idam asya
hanihetur idam asyadhigantavyam idam asyadhigamahetur iti
bhutany anukampante,
'The experts, having directly perceived the dharman [the
true state of things], show compassion on living beings in
the following way: this is [something] to be abandoned,
this is the cause of the abandonment for him [ie for the
living being the expert wishes to instruct], this is
[something] to be attained by him, this is the cause of
the attainment for him (NBh 223.5-7)*. The instruction of
the experts must have practical value for the unenlight-
ened ordinary person who wishes to know what is harmful
for him and what is beneficial. (As regards the term
dharma the reader is referred to the discussion on NBh
40.1-3).
Only practical, valuable instruction given by
33
trustworthy experts can be regarded as reliable statement.
As a means of valid cognition, such a reliable statement
is verifiable for the hearer only in so far as the state-
ment yields the desired result in daily life. This means
that the object described by the reliable speaker must
eventually become perceivable to the hearer. The truth of
a reliable statement depends on the subsequent direct
perception by the hearer of the described object.
Paksilasvamin expresses this view in NBh, p. 25.8-26.2:
sa ceyam pramitih pratyaksapara. iijnasitam artham
aptopadesat pratipadyamano linqadarsanenapi bubhutsate#
lingadarsananumitam ca pratyaksato didrksate, pratyaksata
upalabdhe 'rthe jijnasa nivartate.1 Moreover, [all] this resultant cognition [acquired by in-
ference and reliable statements] has as its ultimate [aim]
the direct perception [of a thing, ie sound inference and
statements are based on direct perception and must ulti-
mately lead to an object of direct perception]. [For] a
person who cognizes through the instructions of an expert
a thing he wishes to knowf seeks to cognize [this thing]
also through seeing a property [of the described thing by
which property he can infer the presence or the existence
of this thing], and then he wants to see through direct
perception the [thing] he has inferred through seeing a
property [of it]. When the [previously described and sub-
sequently inferred] object has been cognized through di-
rect perception, [only then] the seeking to know stops1.
According to Paksilasvamin, the process of cognizing a
thing is only completed with the direct perception of it
if this thing has been first announced to the perceiver by
way of a reliable statement about it. Cognition acquired
through reliable statements and inferences remains in-
decisive as long as it is not corroborated by direct per-
ception. Assertions made by a certain speaker about things
that are not manifest to the senses (adrstartha) can only
be relied upon in so far as the speaker has made correct
assertions on visible things.
drstarthenaptopadesenayurvedenadrstartho vedabhago1numatavyah pramanam iti.
34
'Of the portion of the Veda of which [portion] the object
is not manifest [to the senses], it must be inferred that
it is a means of valid cognition by [analogy with] the
medical texts which are [the same] experts1 instructions
of which the object jjs manifest [to the senses] (NBh
224.5-6)'. Still, whether the assertions of a reliable
speaker are about visible or invisible things, the main
criterion of trustworthiness remains the speaker's exper-
tise and moral superiority manifesting itself as sym-
pathizing with suffering fellow-beings. The sort of in-
ference on reliability that Paksilasvamin alludes to, can
be constructed as follows: (all) expert, compassionate
persons speak the truth, their assertions on visible
things are manifestly true; therefore their assertions on
invisible divine matters (recorded in the Vedas) are also
true.
35
NOTES
1 . Since most of the assertions -- including those on meta-
phycial problems such as final emancipation and the eter-
nity of the soul -- in this school are supported by
reasoned discussion, nyaya, the whole system came to be
called Nyaya, cf Oberhammer 1964a, p. 308 f. The foremost
characteristic of the system was philosophical debate,
epistemology and logic, and it was precisely its episte-
mology and logic that provided a basis for further
developments in that province of learning; cf also
Ruben's remarks in NS (R) p. XIV, '... das war die
geschichtliche Tat des Begrdnders des Nyaya-Systems,
Aksapada Gautama, dass er die Behandlung der Erkenntnis-
mittel vor den GegenstSnden der Erkenntnis als neues
Kapitel der Philosophie festlegte1. And further, NS (*R),
p. XV: 'Aksapada stellt...die Reihe von 12 Erkenntnis-
gegenstSnden auf, die die Seele als Subjekt der Erkennt-
nis, die Sinne usw. als Werkzeuge zur Erkenntnis und den
Erlflsungsweg als Frucht der Erkenntnis behandeln. Aksapada
wurde damit der Begrflnder einer neuen Philosophie, indem
er die Erkenntnismittel, diese Erkenntnisgegensta*nde und
dialektische Begriffe der Debattierkunst in den Kategorien
des Nyaya als Inhalt seines Systems feststelltef.
2. It is very difficult to fix an exact time of compositon for
the NS; Tucci sums up various possibilities ranging from
200 AD to 450 AD, Tucci 1929, p. XXIII-XXIV. The same
dates are repeated in B.K Matilal, Nyaya-Vaisesika,
A History of Indian Literature, Vol VI Fasc. 2, Wiesbaden
1977, p. 78.
3. One only has to compare the Mimamsa Sutra, the Brahma
Sutra and especially the different versions of the
Vaisesika Sutra with the NS to see how much more sys-
tematic the latter is. By the time the final redaction of
the NS took place, these other texts already existed in
some form. It seems that the Vaisesika system of natural
philosophy especially had considerable influence on chap-
36
ters 2-4 of the NS, cf Frauwallner Gesch. II, p. 7, 73 and
Tucci 1929, p. XXVIIf, as well as Oberhammer 1963, p. 70,
And yet chapters 2-4 should be regarded as the rational
inquiry into the validity of the definitions given in
chapter 1 . This inquiry sometimes entails lengthy digres-
sions, so much so that they almost seem to be interpola-
tions. Ruben called NS 2-4 the polemical chapters and
thought some of these chapters might well represent the
contents of the (oral) commentary given by Aksapada on the
definitions of NS 1, NS (R), p. XVI.
4. Frauwallner maintains that Vasubandhu's little manual on
epistemology, logic and dialectics, the Vadavidhi, fol-
lows the traditional arrangement of any such handbook, an
arrangement of which NS 1 and 5 are a classical example,
cf Frauwallner 1957, p. 107, 129 as well as Frauwallner
1959, p. 93.
5. Frauwallner thinks Paksilasvamin wrote his commentary in
the first half of the 5th century AD, Frauwallner Gesch.
II, p. 22; Potter too gives nearly the same time, 425 to
500 AD, Potter 1977, p. 239.
6. The Buddhist Dharmakirti, who wrote centuries later and
whose epistemology and logic goes far beyond the achieve-
ments of the NS and the NBh, has never lost sight of this
practical value of true knowledge; for he opens his NB
with an assumption that is very similar to NS 1.1.1,
samyagjnanapurvika sarvapurusarthasiddhir iti tad
vyutpadyate (NB 1.1),
'The [succesful] accomplishment of every human goal is
preceded by correct [true] knowledge [as a necessary con-
dition for succesful accomplishment], therefore this
[correct knowledge] is [further] explained [in NB] f.
7. It is noteworthy that final emancipation (apavarga) is
considered an object of valid cognition, or rather an ob-
ject deserving to be cognized through the means of valid
cognition, as it is the last of the 12 prameyas (objects
37
of valid cognition) enumerated in NS 1.1.9. This list
also contains such metaphysical objects and notions as the
eternal soul (atman) and the hereafter.
atmasarlrendriyarthabuddhimanahpravrttidosapretyabhavapha-
laduhkhapavargas tu prameyam (NS 1.1.9),
'That which deserves to be cognized [through means of
valid cognition, ie that which is the object of valid
cognition] is [the following:] the soul, the body, the
senses, the objects [of the senses], cognition, the mind,
activity, moral flaws, existence in the hereafter, results
[of good and bad acts], suffering and emancipation [from
suffering]1. Thus, these matters are all capable of being
cognized and rationally inquired into through the means of
valid cognition, the pramanas.
8. Cf Ganganatha Jha's remark in NS (G) vol I, p. 83 (number
at the bottom of the page) 'The Sutra does not mean that
the knowledge of all the categories enumerated is the
direct cause of the attainment of Highest Good; what it
means is that the knowledge of these is conducive to that
end1, hence he translates NS 1.1.1 'it is the knowledge of
the real essence (or true character) of the following
sixteen categories that leads to the attainment of the
Highest Good ' op cit p. 37.
9. Freedom from suffering as the highest goal of Nyaya
reminds us of Buddhism, in which the cessation of suffer-
ing is also an important aim. It is the subject matter of
the third noble truth, duhkhanirodha, 'the cessation of
suffering'.
10. It may seem strange, even contradictory, that NS 1.1.1
says 'true knowledge pj[ or concerning the pramanas 'when
true knowledge is itself the result of pramanas. But this
could be explained thus: although the pramanas produce all
our real knowledge, it is still possible and necessary to
specify our sources of knowledge by determining their num-
ber and defining the functions proper to them. We might go
so far as to say that the pramanas as sources of true
38
knowledge gradually define themselves with increasing
precision. This actually happened in the course of time,
when more precise definitions of the pramanas and their
number were offered by subsequent thinkers as Vasubandhu,
Dignaga, Uddyotakara, Dharmakirti etc. The epistemologi-
cal as well as the logical thinking originating from the
NS and the NBh, became increasingly elaborate with these
authors.
11. Mr H. Krasser in Vienna has drawn my attention to an ar-
ticle by T. Burrow, in which it is argued that the root
ma from which the epistemological term pramana is derived,
should have as its basic meaning "to ascertain" rather
than "to measure" (T. Burrow, Sanskrit M£- 'to ascertain1,
Transactions of the Philological Society, Oxford 1980, p.
134-140). Our present interpretation of pramana as "means
of valid cognition" is based on the following paraphrase
of the term in the NBh:
upalabdhisadhanani pramananiti samakhyanirvacanasamarthyad
boddhavyam (NBh, p. 24.5),
'Pramanas are means (sadhana) [by which to acquire
correct] cognition (upalabdhi); [this meaning of the word
pramana] must be understood on the strength of the etymo-
logical explanation of the term'. The reference to this
NBh-passage I owe to Prof Steinkellner in Vienna.
12. Many, from Paksilasvamin to modern authors, have tried to
interpret the definition in NS 1.1.5, but, it seems, in
vain. The second part of the sentence remains as unintel-
ligible as ever. One only has to compare Oberhammer's
thoughts set forth in Oberhammer 1966, p. 70-71 and
Wezler's reaction thereon in Wezler 1968, p. 198 and his
own explanation brought forward in op cit p. 208-209, with
the older discussions in NS (V), p. 4 and NS (R), p. 3; p.
159-161, note 16. Strangely enough, we find no mention of
all this literature in Chakrabarti 1977, although that
monograph on NS-logic devotes a whole chapter (chapt. 2 to
be exact) to the threefold inference. Chakrabarti just
accepts Paksilasvamin's exposition as if it were per-
39
fectly intelligible and harps upon the third variety,
samanyatodrsta.
13. This simple idea of connexion developed into the logical
theory of positive and negative invariable concomitance,
anvaya and vyatireka, found in later logic. The very
validity of the inference, and hence of the syllogism,
came to be based on invariable concomitance, as will be
shown in the account of Dignaga's and Dharmakirti's logic.
The inference, as presented by Paksilasvamin, constitutes
nevertheless in barest outline the workings of deduction;
the presence of fire, for instance, is deduced from the
presence of smoke which must always be accompanied.by
fire. Using the terminology of Aristotelian logic to for-
mulate the Nyaya-type of deduction, we might say: the
major premiss is that everything that has smoke has fire;
the minor premiss is that at this given place we observe
smoke, so that it follows that this place has fire. Yet
the invariable concomitance of the two objects, necessary
for a valid inference, is not referred to by Paksila-
svamin. He only says
hetudaharanayoh samarthyam paramasuksmam duhkhabodham,fThe efficacy of the reason and the example is very
subtle, difficult to understand (NBh 86.11)'. This means
that we must find out for ourselves whether the reason is
always accompanied by the probandum and whether the ex-
ample actually proves that the probans and the probandum
are connected.
14. "Desire etc." refers of course to NS 1.1.10:
icchadvesaprayatnasukhaduhkhajnanany atmano lingam,
'Desire, aversion, effort, joy, grief and knowledge are
the mark of the soul1. It is interesting to note that the
word linga can also mean logical mark; thus NS 1.1.10
seems to foreshadow Paksilasvamin1s inference on the soul.
15. Unfortunately, the problems of interpretation of NS 1.1.5
and the NBh on it are too numerous to be treated at length
in this context, but in sum we might say that the problems
40
arise from the fact that NS 1.1.5 is no real definition at
all; the process of inference as such does not seem to be
defined, all we find is the definiendum "inference"
without an intelligible definiens. The three terms
purvavad, sesavad and samanyatodrsta have especially
puzzled everyone, even an old commentator like Paksila-
svamin, as is evident from his divergent interpretations.
16. An elaborate treatment of the threefold division of anu-
mana in the NS and the NBh would certainly take up too
much space and go far beyond the scope of this treatise.
For the present, it suffices to summarily recount Paksila-
svamin's own ideas on inference which shaped later epi-
stemological and logical thinking far more than this
single dark statement of NS 1.1.5.
17. Later, we will get acquainted with the criteria of trust-
worthiness as developed by Paksilasvamin in his commentary
on NS 1.1.69. That passage in the commentary is quite im-
portant as it bears upon Dharmaklrti's conception of the
trustworthy speaker, and hence his conception of the Bud-
dha as the ideal reliable preacher of religious doctrines.
18. In Chakrabarti 1977, p. 35, we find this translation of
NS 1.1.34: 'The reason (hetu) is [that from which we get]
what proves the probandum by virtue of the [universal]
concomitance (sadharmya) [stated] in the Exemplification1.
But such universal concomitance, although most probably
implied, is not taught here in the NS nor the NBh. As a
major logical doctrine it was developed later by the Bud-
dhist logicians Vasubandhu and Dignaga. Of course it is
true that probans and probandum must be so connected as to
yield a valid inference and a valid syllogism, but an
elaborate theory of concomitance is not found in the old
Nyaya. Chakrabarti's book -- though containing many useful
observations — is written on the basis of the unfounded
assumption that invariable concomitance is an integral
part of the logical theories in the NS, cf for example op
cit p. 7.
41
Vidyabhusana, on the other hand, translates NS
1.1.34: !The reason is the means for establishing what is
to be established through the homogeneous or affirmative
character of the example1 NS (V) p. 14; the term sadharmya
carries more weight than homogeneity, for it can be inter-
preted as fthe fact of possessing the same property1,
which means possessing the same property as the subject
(sadhya). This 'same1 property acts as the hetu in the
syllogism. The translation of sadharmya with homogeneity
leads to the idea that the example in the syllogism is
simply an analogous instance to make the reason more plau-
sible, thus creating in the mind of the reader the impres-
sion that the NS is propounding a (primitive) proof by
analogy. Ruben renders NS 1.1.34 fDas Beweisen des zu Be-
weisenden durch die Gleichheit seiner Attribute mit denen
des Belegs ist die "Begrtindung", NS (R) p. 11, which is
still inaccurate because of the plurality of 'Attribute1
-- attributes -- where only one property, the probans must
be the ' same' in the subject and the generally accepted
fact, drstanta.
Jha's translation of NS 1.1.34 seems to do full jus-
tice to the Sanskrit: 'The "Statement of the Probans" is
that which demonstrates the Probandum, through its simi-
larity (ie a property common to it and) to the example1,
NS (G) p. 343.
19. Frauwallner always thought that the Nyaya-syllogism con-
stituted a primitive proof by analogy (Frauwallner 1957,
p. 106). Frauwallner 1959, p. 93 says the following: 'in
der Mlteren indischen Dialektik war die Beweisftthrung auf
einen blossen Analogieschluss gegrttndet. Man verwies auf
ein Beispiel [ie drstanta], bei dem der Grund mit der
Folge verbunden erscheint, und folgerte dementsprechend
auch im gegebenen Fall aus dem Vorhandensein des Grundes
das Vorhandensein der Folge [ie sadhya]. Das Gleiche
konnte auch mit Hilfe eines Gegenbeispiels geschehen. Man
zeigte an einem solchen Beispiel, dass im Zusammenhang mit
dem Fehlen des Grundes auch die Folge fehlte, und schloss
daraus, dass im Gegensatz dazu beim Vorhandensein des
42
Grundes die Folge vorhanden sein mttsse. Die Mangelhaftig-
keit dieses Verfahrens scheint uns auf der Hand zu liegen.
Trotzdem war z.B. die Schule des Nyaya bis zur Zeit
Dignaga's nicht Uber diesen Stand hinausgekommen1. In this
passage "Grund" means hetu and "Folge" sadhya, probandum.
Even though Frauwallner does not specify the texts of the
"old Nyaya", he must be referring to both the NS and the
NBh since he says that Nyaya acknowledged only a primitive
proof by analogy up to the time of Dignaga, that, is, in
the time that the NBh already existed. Thus he believes
that the NBh too teaches only proof by analogy. His
opinion about Nyaya-logic was probably influenced by
Ruben's observation in NS (R), p. 11-12; 179 to the effect
that the syllogism taught in the NS is a simple
"Analogieschluss". This interpretation of the syllogism in
NS seems to be based on how one understands the drstanta
which in syllogistic reasoning is the example, udaharana
(drstanta is a thing or fact, udaharana the formulation of
it in the syllogism). If a drstanta is merely interpreted
as one single analogous case of which ordinary people and
learned investigators have the same cognition (NS 1.1.25),
there is justification for saying that the NS (but not the
NBh anymore) teaches a proof by analogy. On the other
hand, if drstanta is interpreted as a fact which is seen
by all to be always the case -- and such an interpretation
of NS 1.1.25 is not impossible — , then we have a real
deduction even in the NS. In the positive drstanta, the
probans is seen together with the probandum; if this is
seen to be the case only once, it is an analogous case;
but if everyone until now has seen the probans to be al-
ways accompanied by the probandum, it is no more one
analogous case but an established, perceived fact. The
root cause of the problem is whether the word "artha" in
NS 1.1.25 refers to one single case or to a fact consist-
ing in an indefinite number of observed cases. It is my
personal opinion that in NS 1.1.25 the latter is the case
(especially when drstanta is used in reasoning), so that
NS 1.1.33-39 can be understood as an (in fact far too) im-
plicit description of a deductive syllogism, ie a form of
43
reasoning in which a particular fact is deduced from an
indefinitely large number of observed similar instances.
However, what seems to be implied in the NS is made ex-
plicit by Paksilasvamin who clearly interprets the posi-
tive drstanta as an indefinite number of cases in which
the probans is seen with the probandum. What is said here
about the positive drstanta equally applies to the con-
trary drstanta. Ruben already has said (contrary to
Frauwallner) that the NBh teaches more than a proof by
analogy. According to him, it teaches the rudiments of de-
ductive reasoning, since the positive drstanta is 'das die
AllgemeingUltigkeit dieses VerhSltnisses [ie of probans
and probandum] belegende Beispiel1, NS (R), p. 179, note
80. The NS offered the basic framework of syllogistic
reasoning in a very implicit manner, while the NBh turned
it into a more elaborate and explicit description of
deductive reasoning, thus laying the groundwork for
Vasubandhu*s and Dignaga's logic.
20. If they would not be, the inference and the syllogism in
which a drstanta showing the connexion of these two
properties is used, would become rather futile. It was
left to Vasubandhu in his description of inference and
the syllogism to state that the probans must be invariably
concomitant with the probandum. Since Vasubandhu, con-
comitance has become an essential part of logical theory,
since it explains why the probans proves the presence of
the probandum.
21. In later logic, we find the socalled threefold (trirupa)
reason (hetu), which teaches that the probans must be
present in the subject (paksa), must be present in the
positive examples (sapaksa) and must be absent from the
contrary examples (vipaksa). These three canons of the
correct reason in inference and the syllogism are fre-
quently attributed to Dignaga or a close predecessor of
his, but they are already present in the NS and clearly
set forth in the NBh. According to the NS and the NBh the
probans must occur in the positive drstanta, never in the
44
contrary drstanta, while the probans is a known property
of the subject, ie occurs in it. The credit for indicating
that the threefold reason is actually found in the NS goes
to Randle who, incidentally, mentions this in Randle Frag,
p. 66, but cf also Randle 1930, p. 366-367. In all fair-
ness, we should add here that Ruben -- whose book was pub-
lished in 1 928 -- says the threefold hetu is an early Bud-
dhist 'Weiterbildung der Lehre des Nyaya, dass der Grund
und das Beispiel positiv und negativ sein ktinnen', NS (R),
p. 179-180.
22. It is possible to convert this into an Aristotelian syllo-
gism; major premiss (example): all produced substances are
non-eternal, minor premiss (reason): all sounds are pro-
duced, conclusion: all sounds are non-eternal. The dif-
ference, apart from the order of the members, is that in
the Indian syllogism it is simply said fsound' without any
quantifier; nevertheless, the idiom of Sanskrit allows us
to interpret 'sound1 as referring to all sounds, sound in
general, as we can say 'man1 meaning all men, not some or
many men. Similarly 'sound' in the Indian syllogism means
every possible sound, not many or some sounds, Chakrabarti
1977, p. 37-38 mentions the fact that in this form of
Indian logic quantifiers are absent, because the proposi-
tions in the syllogism are intended to be only general
propositions.
23. Paksilasvamin does not explain why compassion is part of
the authority (pramanya) of experts. He may, however, have
intended something like the idea that compassion could be
a guarantee for the hearer that the expert -- being com-
passionate and therefore unselfish — correctly communi-
cates his knowledge only in the best interests of his fel-
low beings.
45
II. Buddhist Epistemology and Logic before Dharmaklrti
2.1 Vasubandhu
In order to grasp the importance of Dignaga's contribution
to epistemology and logic in PS and to recognize the
rudimentary ground on which the edifice of Dignaga1s
thinking is built, it is necessary to outline the theories
of his Buddhist predecessor, Vasubandhu. Fragments in
Sanskrit and Tibetan of two of Vasubandhufs works on
dialectics, logic and epistemology have come down to us
and were collected by Frauwallner (in two articles,
Frauwallner 1933 and Frauwallner 1957). These two works
are called Vadavidhana and Vadavidhi, the former seems to
have been a dialectical work dealing with logical proof
(sadhana) and refutation (dusana), while the latter also
sets forth a theory on perception (pratyaksa) and inference
(anumana), and thus contains some epistemological thought.
Vadavidhana and Vadavidhi treat much of the same topics as
the NS and the NBh. According to Vasubandhu there are only
two means of valid cognition -- although the term pramana
is not found in the extant fragments -- namely, perception
and inference. They are the sole means by which we acquire
knowledge for ourselves (cf Frauwallner 1957, p. 120).
Knowledge is communicated in the form of logical proof
(sadhana) consisting of syllogisms. Vasubandhu regards the
syllogism as an explicit inference since he calls the pro-
bandum, ie what is to be proved by the syllogism, rjes su
dpag par bya ba, in Sanskrit anumeya, the thing to be
inferred. This we could render as 'inferendum1 (cf
Frauwallner 1957, p. 135, fragm 3).
2.2 Perception
In the Vadavidhi, perception is defined, tato ' rthad
vij nanaiti pratyaksam, 'Knowledge [that springs] from the
thing [alone] is perception (Frauwallner 1957, p. 138,
fragm 9 ) 1 . This definition is further explained,
yul gan gis rnam par ses pa tha snad bya ba de kho na las
46
gal te de skye zin gzan las ma yin la / de las gzan las
kyan ma yin pa^i ses pa de mnon sum ste / dper na gzugs
la sogs pa^i ses pa dan bde ba la sogs pa^i ses pa bzin
zes pa"o //
'When knowledge comes from [ie is caused by] only that
[real] object [in accordance] with which [this knowledge]
can be designated, and not from another [object], then
this knowledge [which is] not also [caused] by [something]
different from this [first real object] is direct percep-
tion, as the knowledge of [outer objects such as] forms
etc.f and the knowledge of [internal objects such as] hap-
piness etc (Frauwallner 1957, p. 1371, fragm 9, lines 2-
5) 1 . It is important to note that perception is not con-
sidered to be only knowledge received by the senses -- as
in NS 1.1.4 — but perception must also include the intro-
spective awareness of feelings and emotions. Vasubandhu
seems to maintain that it is only after objects have been
directly cognized that we are able to give names to them,
when he says 'When knowledge comes from [ ie is caused by]
only that [real] object [in accordance] with which [this
knowledge] can be designated1. This can be related with NS
1.1.4, where perception is defined as avyapadesya, 'not
[yet] expressible in language1. Both Vasubandhu and the NS
1.1.4 seem to assert that a thing can be given a name only
after it has been perceived by pure sensation. The NS-
definition emphasizes the purely sensational character of
perception before it is put in words. Vasubandhu, on the
other hand, seems to emphasize that only after having per-
ceived an object is one capable of giving it a name. Yet
both say that perception is a form of cognition that may
not be constituted by language.
Vasubandhu further says that his definition is in-
tended to exclude from genuine perception, erroneous
knowledge, ^khrul pa*i ses pa, knowledge of what is con-
ventional or conceptual, kun rdzob pa"i ses pa, and
knowledge derived from inference, rjes su dpag pa"i ses
pa. He says,
*di ni "khrul pa^i ses pa bsal te / dper na na phyis la
dnul gyi ses pa lta bu"o // de ni dnul gyis ses pa"o zes
47
tha snad byed pa la de dnul las skye be van ma yin gyi /
na phyis kho nas de bskyed par bya ba^o //
'With this [definition of perception as knowledge solely
caused by the object], erroneous knowledge is rejected [as
belonging to perception as pramana], as the [erroneous]
knowledge of silver with regard to mother of pearl. While
[in accordance] with silver this [erroneous knowledge] is
designated [by the perceiver as:] 'knowledge of silver1,
it does not also come from [ie is not caused by real]
silver, but it is caused solely by mother of pearl1. This
example is similar to NBh, p. 30,1-3 (translated and dis-
cussed in the previous chapter) where the notion 'water1
attributed to a mirage is rejected as right perception.
Vasubandhu continues,
kun rdzob pa^i ses pa yan *dis gsal te / de ltar ni bum
pa^i ses pa zes pa *di ltar de bum pa la sogs pa rnams
kyis tha snad bya zin / de de rnams las ^byun ba ni ma yin
te / de rnams ni kun rdzob tu yod pa nid kyis rgyu ma yin
pa nid kyi phyir ro // gzugs la sogs pa de ltar yan dag
par zen pa rnams kho na las de ^byun no //
'With this [definition] knowledge of what is conventional
[ie conceptual] is rejected too. Such [conceptual know-
ledge] can thus be designated [as:] "knowledge of a pot"
[in accordance] with [conventional, conceptual objects
such as] pots etc, [but] this [knowledge] does not come
from those [conventional objects] because by their only
existing conventionally [ie as concepts] they are not the
cause [of this conventional, conceptual knowledge]. This
[knowledge] comes only from forms etc that are correctly
ascertained thus [ ie as pots etc]'. Knowledge of objects
(such as pots) that exist only conventionally, that is, as
a concept projected on to a certain form or collection of
parts, is rejected here as true direct perception. Kun
rdzob tu yod pa, samvrtisat, 'what exists conventionally
or conceptually1 is explained by Vasubandhu in the
Abhidharmakosabhasya as follows,
yasminn avayavaso bhinne na tadbuddhir bhavati tat
samvrtisat. tadyatha ghatah. tatra hi kapalaso bhinne
ghatabuddhir na bhavati
48
(AKBh, p. 334.3-4), [When,] after [a thing] has been
divided into [its component] parts, there is no cognition
of the [thing any more], [then] that [thing] exists [only]
conventionally [ ie as a concept]. As a pot [exists only
conventionally], for when the [pot] is divided into pot-
sherds [ie its component parts], there is no cognition of
the pot [any more]1. Lastly, Vasubandhu differentiates
perceptual knowledge from inferential knowledge.
rjes su dpag pa^i ses pa yan ^di kho nas bsal te / du ba"i
ses pa dan ^brel pa dran pa dag las kyafi de byun gi me kho
na las ma yin pas so //
'With the same [pratyaksa definition] inferential know-
ledge is rejected, because, [for example,] this [inferen-
tial knowledge of fire] comes from the knowledge of smoke
[through perception] as well as the recollections of the
[inseparable] connexion [of smoke with fire], but [the in-
ferential knowledge] does not [at all] come only from
fire1 (all three passages from Frauwallner 1957, p. 137,
fragm 9 lines 5-15). Although the last two negations are a
significant improvement on the old NS-definition of per-
ception, the idea from NS 1.1.4 that perception should be
non-erroneous, is retained. The Tibetan ^khrul pa most
probably translates Sanskrit bhranta, 'erroneous1. Now,
whether we say that perception should be avyabhicari (NS
1.1.4) or not bhranta really amounts to the same state-
ment, for vyabhicari and bhranta are synonymous. With some
degree of certainty we may say that Vasubandhu used the NS
1.1.4 and the NBh as one of the starting-points for devel-
oping his own perception theory, which can be summarized
thus: perception is pure awareness of any object whatso-
ever, outer object or internal mental object; this aware-
ness, in order to have the status of means of valid cogni-
tion, must be non-erroneous, must be free from concep-
tualizations and, lastly, must not be the product of
inference.
2.3 Inference
Vasubandhu maintains, in accordance with NS 1.1.5, that
49
inference (anumana) as a means of valid cognition is
preceded by perception, because the act of directly per-
ceiving is mentioned in his definition of inference as the
basis of the inferential thought process.
nantariyakarthadarsanam tadvido ^numanam,
'The observation of an object that is invariably con-
comitant [with another object] for him who knows the
[invariable concomitance of the one object with the other]
constitutes inference (Frauwallner 1957, p. 138, fragm
10)'. If we have seen before through perception and hence
know that two things are always connected, ie are in-
variably concomitant (nantariyaka), we can validly infer
the presence of one object from seeing only the object
that always accompanies it. Vasubandhu gives an example of
invariable concomitance in the commentary on this
definition, dper na me^i du ba lta bu^o, 'for example as
smoke [is invariably concomitant] with fire (Frauwallner
1957, p. 138, fragm 10, line 3 ) 1 . We know, because we have
observed it, that whenever there is smoke, there in-
variably is fire. Only someone who knows that smoke is in-
variably concomitant with fire, ie has seen that smoke is
always produced by fire, can validly infer fire from
seeing smoke alone. This is an improvement of Paksila-
svamin who maintained that inference is 'seeing the con-
nexion between a property and the property-bearer and
seeing the property [connected with the property-bearer]
(NBh 33.3) f, for Vasubandhu explicitly states that the in-
ferendum must be invariably concomitant with the inferens
to generate a correct inference . Paksilasvamin simply
demands that they be connected. Yet, I think it is evident
that Vasubandhu built his theory of inference on Paksila-
vamin's idea of inference found in the NBh passage quoted
above. Both thinkers define inference in general as in-
directly seeing or knowing an object to be present some-
where from actually seeing an object that is connected
with the object that is to be inferred. But to this
general principle, Vasubandhu adds the principle of in-
variable concomitance. Vasubandhu describes inference
proper rather briefly, but he unfolds his theories on
50
deductive reasoning (which is inference after all) fully
in his description of the syllogism. In this he follows
the pattern of the NS and the NBh, where deduction is only
fully set forth in the exposition of the syllogism.
2.4 The Syllogism
Vasubandhu regards the syllogism as the sole means of
soundly proving a fact to others, and hence he seems to
imply that the only way of communicating true knowledge is
through the syllogism. For he does not mention anywhere in
the extant fragments comparison (upamana) and reliable
statement (sabda) as means of valid cognition, nor does
tradition ascribe to him the acceptance of these two
pramanas. His successor Dignaga, for example, nowhere says
that Vasubandhu accepted more than two pramanas, nor does
Uddyotakara in his Nyayavarttika. From their silence on
this, we must assume that the recognition of pratyaksa and
anumana as the only pramanas originated with Vasubandhu.
In Buddhist epistemological thinking after him, the scheme
of two pramanas became authoritative.
Vasubandhu prescribes that logical proof (sadhana)
consists of three members instead of the five of the old
Nyaya (cf Frauwallner 1957, p. 118). The three steps in
Vasu-bandhu's syllogism are, pratij na, thesis, hetu,
reason or probans and drstanta, generally-accepted fact.
We recognize the Nyaya-terminology here, although the
third step is not called udaharana, 'example1 any more but
drstanta, in accordance with the datum that acts as the
real basis of the logical example. Thus the term udaharana
is abolished as inessential.
Since Vadavidhi is Vasubandhufs most important work
on logic (cf Frauwallner 1957, p. 105), and since it is
from that work that the largest fragments have survived
(collected in Frauwallner 1957), I will now give the logic
from Vadavidhi except in the few cases where Vadavidhana
is more explicit.
Vasubandhu first defines the subject in the syl-
logism, pakso vicaranayam isto ^rthah, 'The subject is the
51
thing that is intended for examination (Frauwallner 1957,
p. 135, fragm 1) 1. This examination takes place in the
syllogism, the steps of which are now discussed. In the
thesis one announces that a property, the inferendum, must
occur in the subject, sadhyabhidhanam pratijna, 'The
thesis consists in mentioning what is to be proved [ie the
presence of an inferendum in the subject] (Frauwallner
1957, p. 135, fragm 3 ) 1 . Except for the term abhidhana,
this definition is identical in its wording with the NS
1.1.33, sadhyanirdesah pratijna. Vasubandhu elucidates his
definition by saying,
me dan sa bon dan mi rtag pa nid rnams rjes su dpag par
bya ba nid du dper brjod pa^i phyir chos tsam rjes su dpag
par bya ba nid du mnon par *dod do zes rtogs par bya^o,
'It must be uderstood that only [one] property [of the
subject] is intended to be the inferendum (rjes su dpag
par bya ba, anumeya) [in the thesis] because, for example,
fire, a seed and non-eternity are mentioned as inferendum
(Frauwallner 1957, p. 135 fragm 3) f. In the thesis, one
thing is mentioned of which it is to be proved that it oc-
curs in the subject. What was called sadhya, probandum, in
the NS 1.1.34 is now called anumeya, the thing that is to
be inferred; the latter term could be rendered 'inferen-
dum1. The term sadhya Vasubandhu seems to reserve here for
the relation that is to be proved between the inferendum
and the subject. In the NBh, sadhya could mean^ both the
relation of the probandum with the subject and the proban-
dum as such.3 In the examples of inferenda, we find (a)
fire which must be inferred from the presence of smoke,
(b) a seed which must be inferred to be the origin of a
plant and (c) non-eternity (of sound) which must be in-
ferred from the fact that the subject was produced.
The second step of the syllogism is the reason, ie
that which proves that the inferendum occurs in the sub-
ject, tadrgavinabhavidharmopadarsanam hetuh, 'The reason
consists in pointing to the property which is invariably
concomitant with such [an inferendum like fire, seed, non-
eternity etc.] (Frauwallner 1957, p. 136, fragm 4)'. The
Sankrit term used here for invariable concomitant is
52
avinabhavin, which literally means 'never occurring or
being without1 and hence 'always occurring together'. We
know that smoke always occurs with fire/ so that when we
wish to prove the presence of fire at a particular spot,
we give as a reason the occurrence of smoke at that spot.
Such a reason is valid because smoke always does occur
with fire. After this definition, Vasubandhu says,
don gan sgra mi rtag pa nid la sogs pa bsgrub par bya ba
de "draba ste / de^i rigs can med na don gan zig ^gar yan
^byun ba ma yin pa ste / rtsol ba las byun ba nid ni mi
rtag pa nid lta bu dan du ba ni me lta bu^o zes pa de ni
de *dra ba med na mi byufi ba^i chos can tey
'Such [means:] that thing that is to be proved, [like] the
non-eternity of sound etc. Whatever thing [functioning as
the probans ] does not occur when such [a thing, the in-
ferendum ] is not present, that [thing] is the property
[probans] which is invariably concomitant with such [a
thing, the inferendum], like the fact that production by
effort [is invariably concomitant with] non-eternity, and
like [the fact that] smoke [is invariably concomitant
with] fire (Frauwallner 1975, p. 135, fragm 4)'.
Vasubandhu gives here a definition of invariable
concomitance: the probans is invariably concomitant with
the probandum when the probans occurs with the probandum,
and does not occur when the probandum is not present.
What seems to be missing in this account of the
reason^ is the mentioning of the three criteria of a cor-
rect hetu. It was implicitly present in NS 1.1.34-37 and
NBh^ but Vasubandhu mentions it explicitly in the Vada-
vidhana. The hetu-definition in that work runs, hetur
vipaksad visesah, 'The reason is a thing [ie a property]
that is different from [the properties occurring in] the
[class of objects that are] dissimilar to the subject
(Frauwallner 1933, p. 480, fragm 7 ) 1 . In this rather short
definition, it is said that the reason — being a known
property of the subject (paksa) and occurring in all in-
stances similar to the subject (sapaksa) -- does not occur
in any instances that are contrary to the subject
(vipaksa). In NS 1.1.36-37 and in the NBh, positive in-
53
stances were called positive example, ie the drstanta in
which the probans and the probandum were seen together,
the contrary example being the drstanta in which the
probans never occurs. In the Vadavidhana-definition of the
hetu, the presence of the hetu in the subject and in posi-
tive instances is apparently implied; all that is said is
that the hetu does not occur in the contrary instances.
Fortunately, an important sentence from Vasubandhufs com-
mentary on this definition has survived;
vrttav aha, yo dharmah paksasya sapakse siddho
vipakse nasti,fIn the commentary [Vasubandhu] said, [the hetu is that]
property of the subject which is known [to occur] in the
positive instances and [which] does not occur in the con-
trary instances (Frauwallner 1933, p. 480, fragm 7a)1.
Here it is clearly said that the probans is a property of
the subject (paksa) and it is known to occur in those in-
stances that are similar to the subject, collectively
called sapaksa, while the probans does not occur in the
instances that are dissimilar to the subject, collectively
called vipaksa.
The last step in the syllogism is the generally-
accepted fact, drstanta, from which the validity of the
probans is derived. In the Vadavidhi, the drstanta is
defined as follows: tayoh sambandhanidarsanaiTi drstantah,!The generally-accepted fact [serving as example] consists
in showing the connexion of the two [ ie the invariable
concomitance of the probans with the inferendum]
(Frauwallner 1957, p. 136, fragm 5 ) 1 . The commentary
paraphrases this definition,
de dag gi ste de *dra ba dan de med na mi *byun ba^i
bsgrub par bya ba dan sgrub par byed pa dag gi^brel pa ste
med na mi *byun ba nid gan gis nes par ston pa de dpe^o //
(Frauwallner 1957, p. 136, fragm 5), ' "Of the two"
[means:] of the probandum [ie inferendum] and the probans
that are invariably concomitant, "connexion" [means:] the
invariable concomitance [ ie the never occurring of the
probans without the inferendum], that by which [this con-
nexion] is shown, is called "generally-accepted fact"1.
54
Through the drstanta we have to show the invariable con-
comitance of the probans with the inferendum. This kind of
drstanta corresponds to the notion of sapaksa in the
Vadavidhana-definition of the triple hetuf because the
sapaksa actually evidences that the probans always occurs
with the inferendum. What seems to be missing in Vadavidhi
is the drstanta which shows that the probans never occurs
when the inferendum is absent, ie a drstanta corresponding
to the vipaksa. Only in Vadavidhana is there a definition
of drstanta as directly referring to the triple hetu:
tatha siddho drstantah, 'The generally-accepted fact is
established in this way [ ie as sapaksa and vipaksa ]
(Frauwallner 1933, p. 301, fragm 8 ) 1 . So here we have the
generally-accepted fact showing the positive invariable
concomitance of the probans with the inferendum, and one
that shows their negative invariable concomitance.
Vasubandhu's notion of positive and contrary drstantas can
be compared with Paksilasvamin1s remark on the positive
example, 'By this [ie by this positive drstanta] is ex-
emplified that the two properties function as probandum
and probans (NBh 85.1-2)'; but why the two properties are
probandum and probans he leaves unsaid. About the contrary
example he says, 'This [negative] generally-accepted fact
... does not posses the property [ie the probandum] of the
[subject] because in it [ie in the negative drstanta] does
not occur the property [ ie the probans] of the subject
(NBh 86.2)'. Vasubandhu is much more precise because he
says why two properties are probans and probandum, namely
through positive and negative invariable concomitance,
which comes into view through the two drstantas.
Paksilasvamin1s imprecision arises from his failure to
clearly recognize invariable concomitance as a fundamental
logical principle.
2.5 Dignaga
The most accomplished Buddhist thinker on epistemology and
logic before Dharmaklrti was Dignaga who lived 480-540 AD
(Frauwallner 1961, p. 134-137). Dignaga was a very pro-
55
lific author, as can be seen from the list of works he
wrote (cf Hattori 1968, p. 6-10). In his last work, the
Pramanasamuccaya, "Comprehensive account of the means of
valid cognition", Dignaga sums up his theories on logic
which he had unfolded in various other works and sum-
marizes his epistemological thinking.
By its contents and its systematic treatment of
epistemological and logical problems, the PS exercised
great influence on all subsequent epistemology and logic,
Buddhist and non-Buddhist alike * The PS was, moreover,
the only work of a predecessor that Dharmaklrti deemed
worthy of an extensive discussion in his Pramanavarttika,
that eventually superseded the PS.
2.6 The two Objects of Knowledge
The means of valid cognition, pramanas, are the basis of
all valid knowledge of reality. Dignaga states this prin-
ciple thus, pramanadhino hi prameyadhigamah, 'The
[intellectual] acquisition of the object of valid cogni-
tion is dependent on the means of valid cognition (PS 1.1
commentary line 14) '.8 This was also Paksilasvamin's
starting-point: pramanam antarena narthapratipattih,
'Without a means of valid cognition there is no cognition
of an object (NBh 2.1 ) ' . 9 Now, in the first chapter of the
PS the means of valid cognition are enumerated and their
respective objects defined.
pratyaksam anumanam ca pramane, de gnis kho na ste,
yasmad laksanadvayam prameyam,
'The two means of valid cognition are perception and
inference, only those two because the object of valid cog-
nition possesses two characteristics (PS I.2a-c)'. All
cognizable objects have two appearances and these two ap-
pearances can be known by only two pramartas. In this way
Dignaga derives the number of true pramanas from the two
aspects of reality.10 In the commentary on this verse11,
Dignaga explains the two aspects,
na hi svasamanyalaksanabhyam anyat prameyam asti,
svalaksanavisavam hi pratyaksam samanyalaksanavisayam
56
anumanam,
'For there is no other object of valid cognition than the
characteristic of the [thing] proper and the characteris-
tic of the generality [of a thing]. Perception has as its
object the characteristic of the [thing] proper [ ie the
particular], inference has as its object the characteris-
tic of the generality [ ie the universal] (PS 1.2 commen-
tary lines 20-21 ) f .
The uniqueness (svalaksana) of objects is known
through direct perception (pratyaksa); the general charac-
teristic ( samanyalaksana ) , ie that which several objects
seem to have in common, is that with which inferences
operate. It is possible to translate svalaksana by
'particular' and samanyalaksana by 'universal' as is done
in Hattori 1968, p. 24. We must keep in mind, however,
that they are two aspects of one and the same object, the
real visible aspect which is revealed by perception and
the conceptualized aspect of the thing, which is the field
of inference. The two terms sva- and samanyalaksana were
not invented by Dignaga, but stem from the tradition of
Abhidharma- analysis. A passage from Vasubandhu's12
Abhidharmakosabhasya will give us a better idea of the
terms svalaksana and samanyalaksana.
kayam svasamanyalaksanabhyam parlksate, vedanam
cittam dharmams ca. svabhava evaisam svalaksanam,
'One examines the body in its proper and general charac-
teristics, as well as sensation, mind and dharmas. [Their]
own being is their proper characteristic (AKBh, p. 349.11-
12)'. Their own being, svabhava, evidently refers to the
unique character of every single thing at a given moment
in time. What all things have in common is explained in
the next sentence,
samanyalaksanam tv anityata samskrtanam
duhkhata sasravanam sunyatanatmate sarvadharmanam,fBut the general characteristic is the non-eternity of
produced [things], the fact that [everything that is] con-
nected with the [four] evil influences is [full of] suf-
fering, and the fact that all things are empty and not the
self (AKBh 349.12-13) '. If this is taken in a purely onto-
57
logical sense, we could say that everything is ephemeral,
impermanent; that is to say that everything has its own
unique being which lasts only for a moment. This consti-
tutes the svalaksana. The general characteristics of all
things are their impermanence, painfulness, emptiness and
unsubstantiality; these marks all things have in common
and hence these marks are called samanyalaksana,
The real thing lasts only for one moment, ksana;
therefore everything is said to be momentary. The doctrine
of momentariness is described in Abhidharmakosa IV, verse
2cd and the commentary thereon, na gatir yasmat samskrtam
ksanikam, '[Bodily activity] is not a [flowing] movement
because [whatever is] produced is momentary [ie lasts for
one moment]1. The commentary explains the notion of1 moment'.
ko ^yam ksano nama, atmalabho "nantaravinasi, so 'syastlti
ksanikam. dandikavat. sarvam hi samskrtam atmalabhad
urdhvam na bhavatiti yatraiva jatam tatraiva dhvasyate,
'What is this moment ? [It is] the coming into being which
ends immediately; momentary [means:] for this [particular
thing] there is this [one moment of existence only;]
therefore [the particular thing is called] momentary, as
[we can say] 'having a stick1 [in like manner we can say
having a moment, existing for one moment, in brief 'momen-
tary']. For everything that is produced does not exist
anymore after its coming into being, [that is to say]
where [something] is born, there it decays (AKBh, p.
163.1-2)1.
As all real things (which are the object of percep-
tion in Dignaga's opinion) are momentary, they do not
remain the same at every instant. From this it follows
that the general characteristics, which have to remain un-
varying in order to be "perceptible" of many individual
things, are not really present in the things themselves,
but are conceptualized by the perceiver. Therefore,
Dignaga says, that the general characteristics, the uni-
versals, are the domain of inference, for the drawing of
an inference is a purely mental activity involving the
rearrangement of concepts. Such, at least, would be the
58
drift of Dignaga's argument in PS 1.2 and commentary.
2.7 Perception
Dignaga defines perception in general as that form of cog-
nition which is free from conceptualizing, pratyaksam
kalpanapodham, 'Perception is [cognition] removed from
conceptualizing (PS 1.3c)1. This is paraphrased, ses pa
gan la rtog pa med pa de ni mnon sum mo, 'Perception is
that knowledge in which there is no conceptualizing (PS
1.3c commentary line 6 ) 1 . In PS 1.3d conceptualizing it-
self is defined as follows, namaj atyadiyoj ana,'[It is] the
application of a proper name, a common noun etc [to a per-
ceived object]'. Dignaga regards conceptualizing as the
mental process of giving names to the impressions of
reality received by perception. In the commentary on this
verse, Dignaga distinguishes five kinds of name-giving or
verbal designation. The first is yadrcchasabda, a proper
name denoting only one object, as the proper name
'Pittha', which designates only one person. The second is
jatisabda, a word denoting a class or genus, like the word
'cow*. The third is qurtasabda, a word denoting a quality,
like the word 'white'. The fourth is kriyasabda, action
noun or verb, like 'cooking' and the last is dravyasabda,
a word denoting a substance like dandin, 'staff-bearer' or
visartin, 'horn-bearer', 'horned' (all examples from
Hattori 1968, p. 25, 83-85).
According to Dignaga, direct perception as a means of
valid cognition should be free from all name-giving, it
should not be influenced by our linguistic consciousness.
This reminds us of NS 1.1.4 where pratyaksa has been
defined as avyapadesya, 'not [yet] expressible in lan-
guage', which means that the direct perception of a thing
should not be influenced by language. And furthermore,
Dignaga's definition is reminiscent of Vasubandhu's
perception-definition in Vadavidhi fragment 9. Dignaga's
general definition of perception in PS I.3c seems to sum-
marize and improve on the ones offered by the NS, the NBh
and Vasubandhu.
59
As for perception/ it includes all forms of inner
awareness, feelings etcr so that in this respect also,
Dignaga follows his predecessor Vasubandhu. Dignaga dis-
tinguishes four kinds of perception: sensory awareness,
mental awareness of the sensory impressions, mental aware-
ness of emotions and lastly, the higher perceptions of the
yogins. The object of sensory perception is described
thus:
dharmino 'nekarupasya nendriyat sarvatha gatih /
svasamvedyam anirdesyam rupam indriyagocarah //
(PS 1.5)
'The sensory cognition of a property-bearer [ie a real,
outer object] whose forms [ie appearance] are many is not
[possible] in a complete manner, [but] [that outer] form
[of the particular] which is to be cognized by the self
[-awareness of cognition] and [which form] is not express-
ible in language, [that form] is the object of the [five]
senses.'
The object of pure sensory perception cannot be de-
scribed in language,which means that it cannot be concep-
tualized, but must simply be perceived as it appears to
the senses. In so far as the perceiver does not concep-
tualize the outer phenomenon, he is perceiving an object
of sensory perception. The basis of perception is the fact
that cognition is aware of its own content and does not
need another internal witness for becoming aware of the
knowledge received through the senses. fThe word "sva-
samvitti" (self-cognition: sva-samvid, -samvedana, atma-)
is expressive of the thought that cognition is cognized by
itself and does not need another cognition to cognize it-
self (Hattori 1968, p. 101 ) f . Dignaga himself says,
ses pa ni gnis su snan bar skyes te, ran gi snan ba dan
yul gyi snan ba"o. snan ba de gnis la gan ran rig pa de ni
^bras bur ^gyur ro
(Hattori 1968/ p. 183.18-19)/ '[All] knowledge has come
about with a double appearance, [namely] the appearance of
itself [ie as awareness of itself] and the appearance of
the object [to be cognized]. The awareness of itself [of
knowledge] in this double appearance becomes [in the
60
process of cognizing an object] the effect [ie the resul-
tant cognition through a pramana of an object]1.
Three other types of pratyaksa are described in the
PS 1.6,
manasam cartharagadisvasamvittir akalpika /
yoginam gurunirdesavyatibhinnarthamatradrk //fAnd mental [direct perception] is [, on the one hand,]
the [mental] awareness of itself of an [outer] object
[appearing in the mind, after the object has been per-
ceived with the senses], and [on the other hand, the
awareness of itself of emotions like] passion etc. [Both
forms of mental perception are] without conceptualization.
[The perception] of the yogins is [their] seeing an object
simply as it is, completely dissociated from any descrip-
tion [made of it] by the spiritual preceptor13'.
As we have seen, Vasubandhu opposed erroneous percep-
tion, conceptual knowledge and inferential knowledge to
genuine perception as a means of valid cognition. In this
rejection, Dignaga follows him (as is noticed by Hattori,
cf his book p. 95-96).
bhrantisamvrtisaiinanam anumananumanikam /
smartabhilasikam ceti pratyaksabham sataimiram //
(PS I.7cd-8ab)
'Cognition (a) [consisting] in error and (b) of that which
exists [only] conventionally, (c) [cognition consisting
in] inference and (d) [the resultant cognition] derived
from inference, (e) [cognition] relating to memory and (f)
to desire, together with (g) [cognition blurred by] an
eye-disease, [all these cognitions] have the [false] ap-
pearance of perception'.
In the prose passage following on this verse, Dignaga
explains only why cognitions a-d are not instances of true
direct perception.
tatra bhrantijnanam mrgatrsnikadisu toyadikalpana-
pravrttatvat pratyaksabhasam. samvrtisajjnanam
samvrtisatsv arthantaradhyaropat tadrupakalpanapra-
vrttatvat pratyaksabhasam
(PS I.8ab, comiri line 10-11). 'Concerning this [list of
seeming perceptions], erroneous cognition [bears] upon
61
[real objects that are liable to deceive the perceiver,
such as] mirages etc [and] has the [false] appearance of
[true] perception because [erroneous cognition] has become
active through the conceptualization [of mirages etc] as
water etc. Cognition of what exists [only] conventionally
[bears] upon [things] that exist [only] conventionally
[and it] has the [false] appearance of [true] perception
because [this cognition comes into existence through]
[mental or conceptual] superimposition [on a real thing]
of a thing that is different [from the real thing], [that
is to say:] because [this sort of cognition] has become
active through the conceptualization of the form of the
[conventionally existing thing out of the perception of
things that exist in the ultimate sense]'.13a
In this rather intricate passage Dignaga makes clear
that bhrantijnana, erroneous cognition, cognition consist-
ing in error, and samvrti salj nana, cognition of what ex-
ists [only] conventionally [ie as a concept, not as a real
thing], cannot be considered to be true perception since
these two forms of cognition are not free from conceptua-
lizations, which true perception should be free from
(according to the definition in PS I.3c). Erroneous cogni-
tion occurs when, for instance, someone perceives a mirage
and wrongly forms in his mind the notion of water about
this mirage. Consequently, his "perception" of water is
false. The same example of false perception is given in
NBh 30.1-3 in relation to the postulation that pratyaksa
should be avyabhicarin, non-erroneous.
The second type of false perception corresponds to
Vasubandhu's kun rdzob pa^i ses pa, saittvrtil nana, which is
rejected as true perception in Vadavidhi fragm 9. Accord-
ing to Dignaga, samrtisajjnana operates through the con-
ceptualization of the form of the real thing into a thing
that exists only conventionally, ie by convention or as a
concept. For instance, the form of a certain number of
atoms (which exist in the ultimate sense) is conceptual-
ized as a pot which exists only conventionally (according
to AKBh, p. 334.3-4 and Vadavidhi fragm 9, both quoted
previously on p. 51-52).
62
The conceptualization of a heap of atoms as a pot
consists in conceptually superimposing the notion "pot" on
this heap. Since conceptualization comes into play in the
"perception" of a pot, this perception is denied the sta-
tus of real perception. The difference between the two
forms of false perception lies in the fact that bhranti-
jnana is the result of mentally mixing up really existing
things, eg water and a mirage, while the second type con-
sists in superimposing a conventionally existing thing on
real things.
anumanatatphaladii nanam- purvanubhutakalpanayeti na
pratyaksam,
'Knowledge [as the process] of inference and [knowledge
as] the result of that [ ie inference] etc is not percep-
tion, for [all those forms of knowledge are produced] by
the conceptualization of what was experienced before
[during actual direct perception] (PS I.8a-b), comm line
12) f. Here it is said that inference follows upon ex-
periences received from perception, and as inferential
cognition deals with concepts (which are, after all,
abstractions of real objects), it is excluded from per-
ception. We will remember that Vasubandhu too, excluded
inferential knowledge from perception. Dignaga rejects the
same three forms of seeming perception as Vasubandhu in
Vadavidhi.
2.8 Pramana and its Result are not Different
In the first chapter of the PS, Dignaga discusses an im-
portant theory of his, namely the theory that a means of
cognition and its resultant cognition are in reality not
different. He states, savyaparapratltatvat pramanam phalam
eva sat '[We say that] a means of valid cognition is iden-
tical with the resultant [cognition and not different from
it] because [the resultant cognition] is cognized together
with the activity [of cognizing a thing through a means of
valid cognition] (PS I.8cd)f. Dignaga does not distinguish
between the act of correctly cognizing an object and the
state of having correctly cognized, the state of possess-
63
ing valid cognition of an object. He evidently regards
pramana as the process of knowing a thing and finds it un-
necessary to regard means and result as separate facts.
Such a distinction between the means and the result of
valid cognition was indeed made by Paksilasvamin who said,
sa yenartham praminoti tat pramanam, 'A means of valid
cognition is that by which he [who wishes to cognize a
thing,] validly cognizes the thing (NBh 3.5)'. And he
continues, yad arthavijnanam sa pramitih, 'The valid
cognition is the [true] understanding of the thing (NBh
3.6)'. It seems that Dignaga had this passage in mind when
he wrote in his commentary on PS I.8c-d the following, *di
la phyi rol pa rnams kyi bzin du tshad ma las ^bras bu don
gzan du gyur pa ni med, 'As regards this [statement in PS
I.8c-d we add that we do] not [think] that the result [ ie
the resultant cognition coming from a pramana] is a thing
that is different from the means of valid cognition [it-
self] as the outsiders [ie the non-Buddhist philosophers
maintain] (Commentary line 15, p. 182.8-9; p. 183.9-10)'.
Hattori thinks that word 'outsiders' possibly refers to
the Mlmamsakas, although primarily it seems to refer to
Paksilasvamin (Hattori 1968, p. 99). Dignaga continues,
kyi / ^bras bur gyur pa^i ses pa de nid yul gyi mam pa
can du skyes pa dan / bya ba dan bcas par rtog pa de ne
bar blafis nas tshad ma nid du ^dogs pa ste / bya ba med
par van yin no //
'But only that knowledge that is the resulting [cogni-
tion], has originated while possessing the form of the
[outer] object, and when this [resulting] cognition has
been appropriated to oneself as including the activity [ie
the process of cognizing], then it [ie the activity toge-
ther with the result] is named "means of valid cogni-
tion", although [in fact pramana] is without activity
(comm line 15-17; p. 182.9-12; p. 183.10-13)'. A passage
from Hattori's commentatorial notes may be quoted here to
throw more light on the ideas contained in these words of
Dignaga. 'in asserting the identity of pramanaphala and
pramana in this verse, Dignaga is basing his thought upon
sakarajnanavada, the theory that the cognition as pramana-
64
phala is the apprehension of an object (visayadhigati).
If, ... the cognition were formless (nirakara) while the
object had form (akara), then the cognition itself . .., as
distinguished from the object, would remain the same
whether it cognized something blue or something yellow or
any other object. Accordingly, the cognition as the ap-
prehension of an object must be admitted to be sakara. ...
The sakara cognition is thus understood to possess the
function (vyapara) of assuming the form of an object
(Hattori 1968, p. 98) . Translating vyapara by 'act1 or
'activity1 seems to me to render the term in a better way,
since vyapara refers to the process, the act of cognition,
rather than to the function of assuming.
2.9 Inference for Oneself
The second means of valid cognition is inference (anumana)
which, according to Dignaga, manifests itself in two ways.
The first is svarthanumana, inference for oneself, the in-
ference that is drawn in the mind but not expressed in
words; the second is pararthanumana y inference for the
sake of others, inference verbalized in the form of a cor-
rect syllogism. The first verse of PS II says:
anumanam dvidha svartham trirupal lingato ^rthadrk
(Kitagawa 1968, p. 74)
'inference is twofold. [Inference] for oneself is seeing
an object through a property (linga) which has three as-
pects'. The commentary says about the two forms of infe-
rence, ran gi don daft gzan gyi don dan (Kitagawa 1965, p.
447.2). 'For one's own sake, and for the sake of others'.
The three aspects of the property refer, of course, to the
trirupahetu which was formulated by Vasubandhu in Vadavi-
dhana (Frauwallner 1933, p. 480 fragm 7a). But Dignaga
took the definition of the threefold hetu to also define
the property that we use to infer the presence of another
property, the inferendum. The inferring property could be
called 'inferens1, which renders the term linga used by
Dignaga, a term which, incidentally, is used by Paksila-
svamin to denote the proving property in his exposition of
65
inference in NBh 33*3-5. The triplicity of the linga is
defined,
anumeye 'tha tattulye sadbhavo nastitasati
(PS II.5c-df Randle, Frag p. 22)
'The [linga, property,] is really present in the inferen-
dum and in [the set of instances that is] resembling the
[inferendum], and [the iin^a] is not present when [the in-
ferendum is] not present1. The inferens-property can only
validly prove the presence of the inf erendum if (a) the
inferens occurs in the inf erendum (b) occurs in the in-
stances that are like the inferendum, instances where the
inferendum and the inferens are seen together, and (c) if
the inferens never occurs in instances where the inferen-
dum does not occur. The inferendum itself is defined in
the commentary,
rjes su dpag par bya ba ni chos kyi khyad par du byas pa^i
chos can no
(Kitagawa 1965, p. 455.11-12), 'The inferendum is a
property-bearer (chos can, dharmin) which is distinguished
by a [special] property1. PS II.5c-d itself seems merely
to epitomize what was said in NS 1.1.34-35 and the NBh
thereon, as well as the Vadavidhana-passage mentioned on
page 53. If these three criteria are not met, the in-
ference will not be valid, says Dignaga.
ekaikadvidvirupad va lingan narthah krto 'rthatah
(unpublished fragm found by Steinkellner in Tarkasopana p.
286.5) 'Through a property that has [in] each [case] one,
or [in] each [case] two characteristics, a thing [the
inferendum] is not established [inferred] from the [other]
thing [from the property that proves, the inferens] (PS
II.6c-d)'. There are three cases in which only one cri-
terion is met, and they are: (a) when the property is not
present in the inferendum, not present in similar in-
stances and not present in contrary instances; (b) when
the property is not present in the inferendum, is present
in similar instances as well as in contrary instances; (c)
when the property is present in the inferendum, absent
from similar instances, but present in contrary instances.
As for the three cases in which two criteria are met, they
66
are as follows; (d) when the property is present in the
inferendum, in similar instances and in contrary instan-
ces; (e) when the property is present in the inf erendum,
absent from similar instances and absent from contrary
instances; (f) when the property is absent from the
inferendum, present in similar instances and not present
in contrary instances (cf Kitagawa 1965, p. 456-457 and
Hayes 1980, p. 253-254). There is a last possibility that
the property is absent from the inferendum, absent from
similar instances, but present in contrary instances; in
this case no criterion is met. From the Vadavidhi we have
learned that another important condition for an inference
to be valid is the invariable concomitance of the inferens
with the inferendum. This notion reappears in PS II.
linqasyavyabhicaras tu dharmenanyatra darsyate /
tatra prasiddham tadyuktam dharminam gamayisyati //
(PS 11.11, source PVSV (Gn) p. 95.8-9)
'The invariable concomitance (avyabhicara) of the
[proving] property with the property [ie the inferendum]
is pointed out with reference to another [place where the
invariable concomitance was really observed]. [If the in-
variable concomitance is remembered, then the proving
property] that is known [to be present] there [in the
subject] will indicate [the presence of] the property-
bearer [the inferendum] which is [invariably] connected
with the [proving property, the inferens]1. In the commen-
tary on this verse, Dignaga adduces the well-known proof
of fire from smoke.
gan me dan dud pa med na mi rbyun ba^i ^brel ba de gzan du
bstan par byas nas yul gzan du dud pa de *ba yzig tsam
mthon van gan na dud pa yod pa de na me yod do zes me dan
ldan par grub pa ston par byed do
(Kitagawa 1965, p. 461.5-9). 'When the connexion, that is,
the invariable concomitance of smoke with fire has been
pointed out with reference to another [place where the
connexion was really observed], then also by seeing smoke
alone at another place [different from the instance where
the connexion was seen] is pointed out the proof that
[this second place] possesses fire, since [we remember the
67
fact observed earlier that] wherever there is smoke there
is fire f.
The power of invariable concomitance in a deduction
is, according to Dignaga, quite limited. It will reveal
the presence of a thing in a certain locus, but not its
specific qualities. The invariable concomitant inferens
can reveal the inferendum in general, not the particular
properties of the inferendum.
rtags can gan rjes *brel pa yi / rtags can de *dir
rtogs bya yin / bye brag rtogs bya ma yin te / de dag
nid las ^khrul phyir ro // tasyaiva vyabhicaritvat //
(PS II.T8, Kitagawa p. 465. 9-12; DvNC II, p. 675.11)
'Those property-bearers [inferenda] that are connected
[with a property, the inferens,] can be made known by way
of the property [the inferens], but the specific qualities
[of the property-bearers] cannot be made known [through
the inferens, because] they are not [always] invariably
concomitant with that [ ie with the inferens] alone1. In
the inferendum may occur a wealth of properties, but since
those are not all invariably connected with the inferens,
the latter cannot make them known. The inferens only dis-
closes a general notion of the inferendum, nothing more.
For the particular characteristics of the inferendum are
exclusively known through direct perception.
Dignaga discovered that inseparable connexion is not
sufficient to make a deduction valid. In the case of fire
and smoke, it is true that smoke can reveal the presence
of fire in general on the grounds that smoke is always
connected with fire. But the converse is not always true.
Fire does not necessarily indicate the presence of smoke,
for there are instances of fire without smoke, but no in-
stances of smoke without fire. So it is said that the ex-
tension of fire is greater than that of smoke. All the in-
stances of fire or instances of smoke or of anything else
can collectively be called a set or a class as in Oberham-
mer 1964, p. 136. Thus we can speak of fire as the class
of fiery objects and we can call smoke the class of all
instances of smoke. Now the class of fire has a greater
extension than that of smoke and so it is said that fire
68
pervades (vyapin) smoke and that smoke is pervaded
(vyaptar) by fire. The class with the smaller extension
— being pervaded by a class with a greater extension --
can serve as inferens for the latter, provided the two
classes are inseparably connected. The class which acts as
inferendum must have an equal or greater extension than
the class which serves as inferens. It seems that Dignaga
was the first to work out. the logical theory of pervasion
(vyapti) (cf Oberhammer 1964, p. 136) a theory still
unknown to Vasubandhu.
The inseparable connexion of two classes acting as
inferens and inferendum is explained as the relation of
the content to its receptacle.
sambandho yady api dvisthah sahabhavyangalinginoh /
adharadheyavad vrttis tasya samyogivan na tu //
(PS 11.20, Katsura 1975, p. 67)
'Although the connexion of the concomitant (sahabhavin)
part [the inferens] with the property-bearer [the
inferendum] is relying on the two [ ie on inferens and
inferendum], the occurrence of this [connexion] is like
that of a receptacle and a deposit but not like that of a
thing possessing full conjunction [between the two]'. The
situation of receptacle and what is to be deposited in it
is intended by Dignaga to be such that wherever what is to
be deposited is placed, it necessarily is always placed in
a receptacle, but the receptacle itself can occur without
any content. The receptacle must be as large as, or larger
than, what is to be placed in it. Likewise, the extension
of the inferendum is as great or greater than that of the
inferens. The inferendum can occur without the inferens,
but the inferens, on the other hand, must always invari-
ably occur with the inferendum. Yet inferens and inferen-
dum are not inseparably connected in such a way that they
are similar or equal. Therefore Dignaga compares the rela-
tion between inferens and inferendum to the relation of
the content with its container. In the commentary he
stresses that inferens and inferendum are qualitatively
different from each other in spite of their invariable
concomitance.
69
na kadacid adhara adheyadharma bhavati napy adheya
adharadharma tatha na kadacil lingam lingi bhavati
lingi va lifigam
(PS 11.20 comm, source DvNC II, p. 679.14-15). 'The recep-
tacle never possesses the properties of what is to be
deposited, and also, what is to be deposited never pos-
sesses the properties of the receptacle; likewise, the
property [inferens] is never [the same as] the property-
bearer [the inferendum] or the property-bearer [the same
as] the property1. On the remark in the last quarter-verse
(PS II.20d), the commentary says,
samyogi tu yathaikas tatha dvitlya iti na tadvad iha,
'But a thing possessing full conjunction [with the other
thing] means "as the first thing is, so is the second
thing". Thus it is not here [in this doctrine of recep-
tacle and content symbolizing the invariable concomitance
of inferens with inferendum] (PS 11.20 comm, source DvNC
II, p. 679.15)'. This means that inferens and inferendum
may not be so connected as to be virtually the same, be-
cause in that case the inference would be an instance of
direct perception. For example, fire is not smoke and
smoke not fire; smoke has got properties that fire does
not have and vice versa. We infer fire from smoke, but if
the two were the same, seeing smoke would amount to seeing
fire.
The theory of pervasion and extension is summarized
in PS 11.25.
rva can nid kyis ba Ian khyab /
rva can min pa ldog par byed /
ba Ian rva can nid kyis khyab /
ba Ian min don ldog byed min //
(Kitagawa, p. 469.7-10).
'[The fact of being a] cow is pervaded by the fact of
having horns, [and this fact of being horned] is contrary
to [the fact of] not having horns. [But the fact that] a
cow is pervaded by the fact of having horns is not con-
trary to [the class of] things [ie animals] that are not-
cows'. The extension of the class of horned animals is
much greater than the extension of the class of cows. The
70
property of having horns is, therefore, capable of remov-
ing the notion of the class of unhorned animals, but it
cannot prove an animal to be a cow. Eg when I see a horned
animal, I do not necessarily see a cow, for I may be
seeing some other horned animal; on the other hand, I can
be quite sure that I am not seeing a dog or a rabbit —
animals that have no horns. Conversely, if I see a cow, I
will also see the horns. Thus, having horns negates every-
thing without horns, but does not imply belonging to the
class of cows, whereas belonging to the class of cows
necessarily implies being horned. In an earlier statement
in prose, Dignaga maintains the same,
visanitvena govyapitve fpi na qoprakasanam vyapitvat
tu tad eva gotvena prakasyam bhavati
(PS 11.22 comm, source DvNC II, p. 679.22). 'The fact of
having horns does not reveal [the fact of being a] cow,
although [having horns] pervades [the fact of being a]
cow; on the contrary, only this [fact of having horns],
since it pervades [cows], can be revealed through the fact
of [being a] cow1. It is not possible to infer cowhood
from being horned, although the latter pervades the for-
mer, which means that the two are in part invariably
concomitant; but it is indeed possible to infer horned-
ness from the fact of being a cow. If two classes of ob-
jects are invariably concomitant, it is the class with the
smaller extension alone that can be used as a valid reason
to infer the presence of the class with the greater exten-
sion, but the converse will not always yield a valid infe-
rence. With this theory of pervasion, Dignaga was able to
refine the possibilities of valid inference far beyond the
old Nyaya and Vasubandhu.
2,10 The Syllogism
In PS II we are already confronted with an important part
of Dignaga's logic, albeit not in the formalized ap-
pearance of the syllogism.15 In PS III.1a-b the syllogism
is defined in general, and later in detail. Verbalized
deductive reasoning is now called pararthanumana, in-
71
ference for the sake of others. In so far as the syllogism
embodies a valid inference, it is to be regarded as a
means of valid cognition. PS III.1a-b runs,
parartham anumanaiti tu svadrstarthaprakasanam
'Now, inference for the sake of others is [an instance of]
communicating a thing that one has seen [ie has inferred]
for oneself (source Frauwallner 1957a, p. 60)'. A little
further in the text of the commentary on this verse, we
find this definition, trirupalingakhyanam pararthanumanam,
'Inference for the sake of others consists in communicat-
ing the property that has three characteristics (source
PVBh, p. 468.1; NB III.1 unpublished fragm found by
Steinkellner)'. The threefold property refers to the in-
ferens in the internal inference. When that property is
communicated to others, it becomes the reason (hetu) in
the syllogism.
That step in syllogistic reasoning which the NS, the
NBh and Vasubandhu call pratiina, thesis, is described by
Dignaga in the following way:
tatranumeyanirdeso hetvarthavisayo matah
(PS III.1c-d, source Frauwallner 1957a, p. 60) 'With
regard to this [inference for the sake of others, we]
regard the [verbal] indication of the inferendum [, the
property of which we wish to prove that it occurs in the
property-bearer, ] [not as a necessary part of the syl-
logistic proof itself, but] as [a proposition] referring
(visaya) to the purpose (artha) of the probans [, namely
to prove the presence of the inferendum in the property-
bearer in which the probans occurs]'.
The indication of the inferendum, ie the "thesis"
has, up to Dignaga's time, been regarded as a necessary
part of the logical proof (sadhana). Dignaga seems to say
here that the proof of the presence of an inferendum is
solely done by the probans (a particular proposition) and
the drstanta (a general proposition). The thesis is not an
integral part of the proof, but simply meant to make clear
the purpose of the probans. The inferendum is proved to be
present in a property-bearer. The latter is described in
the prose on PS III.2b-d,
72
... sgrub par 'dod pa^i chos kyi khyad par du byas
pa"i chos can
(Kitagawa 1965, p. 472, text V.4), '...the property-bearer
[is] characterized by a [visible] property [ie the probans
that resides in it], which is intended for proving [that
another property occurs in it]1. A few lines below, the
formulation of the property-bearer characterized by a
property is called a bsgrub bya bstan pa, sadhyanirdesa
(op cit, text K, p. 472.7), fThe indication of what is to
be proved (sadhya) f . This same term is also used in NS
1.1.33, the NBh on it and Vadavidhi frag 3. According to
PS III.10, sadhya can refer to the property whose presence
is to be inferred, the property-bearer in which it is to
occur and the fact that the inferendum occurs in the
property-bearer.
samudayasya sadhyatvad dharmamatre 'tha dharmini /
amukhye 'py ekadesatvat sadhyatvam upacaryate //
(source PVBh, p. 580.27; Steinkellner 1967 II, p. 83 where
it is quoted as PS III.9).
'Since what is to be proved (sadhya) is [really] the ag-
gregate [of the property that is to be proved to occur in
the property-bearer together with the property-bearer
itself], the [term:] "being what is to be proved" is meta-
phorically used [to designate] the property [ ie the in-
ferendum] alone and then the property-bearer [in which the
inferendum should occur] as well, although [they are] not
directly [designated], because [both, ie inferendum and
property-bearer] are [each] an [integral] part of [of the
aggregate]'.
A correct sadhya, ie a property-bearer of which some
property must be proved to occur in it, is described in PS
III.2,
svarupenaiva nirdesyah svayam isto 'nirakrtah /
pratyaksarthanumanaptaprasiddhena svadharmini //
(source Frauwallner 1957a, p. 60)
['What is to be proved, sadhya] must be specified through
its own appearance alone; it [should be] accepted [to
exist] by [the speaker] himself and [the sadhya should]
not [be] contradicted by [another property that is] estab-
73
lished [to occur] in his own [ ie accepted by the speaker]
property-bearer through an object of perception,through
inference or through [the words of] an expert1. The com-
mentary says,
ran gi no bo kho na zes bya ba ni bsgrub par bya
ba"i no bos yin
(Kitagawa 1965, p. 471 text V.11), 'Through its own ap-
pearance alone means: through its appearance as sadhyaf.
And further, the prose text on PS III.2b-d says,
gan sgrub par *dod pa'i chos kyi khyad par du byas pa'i
chos can
ba'i mnon
qrags pa
na bsgrub
[qal
sum
rnams
par
te]
dan
te
bya
de
/
/
ba
la
rjes
chos
kha
bsgrub par
su dpag pa
gzan gyis
na ma tho
bya ba'
dan /
ma bsal
ba med
i chos dan 'gal
lun
ba'
par
dan /
'o // de ltar
bstan pa yin
no //
(Kitagawa 1965, text V, p. 472,4-8), '[when] a property-
bearer which is intended for proving as being charac-
terized by a property, is not removed by another property
that is (a) contrary to the property that is to be proved
[to occur in the property-bearer], [and] (b) is estab-
lished [to occur] in the [property-bearer] through
perception, inference and tradition, [then] in- this way
the indication of what is to be proved is irreproachable1.
The speaker of the syllogism must personally accept
the existence of the property-bearer. That the speaker
really saw the property-bearer for himself was expressed
in the initial phrase of PS III, 'Inference for the sake
of others is communicating a thing that one has seen for
oneself1. It should not be possible to deny the existence
of the property-bearer on the grounds that it really and
observably possesses properties contrary to the properties
it is proposed to have, eg when a speaker made the propo-
sition that fire is cold, for it is established by sense
experience that fire is not cold but possesses the pro-
perty heat. Dignaga mentions the pramanas by which one
could deny the existence of a subject. They are percep-
tion, inference and (remarkably enough) reliable state-
ment, the statement of an apta, an expert, by which
Dignaga is probably referring to the syllogism, although
74
in the commentary he calls it lun, agama, tradition. In
the commentary, he also gives a few examples of false
propositions, ie propositions invalidated by pramanas,
asravanah sabdo nityo ghata iti (PVBh, p. 454.8), '[every]
sound is inaudible, [every] pot is eternal1. The first ex-
ample is contradicted by direct experience since we do
hear sounds if they are loud enough; it is not the case
that no sound can be heard at all. The second example is
contradicted by sense experience as well as inference. We
may see a pot being made or being destroyed or both and
hence deduce its perishableness.
As with internal inference, we need a proving
property by the presence of which we can establish the
presence of another invariably concomitant property that
is as yet not immediately visible in a given property-
bearer. The triplicity of the linga described in PS II.5c-
d reappears here as the triplicity of the hetu in the
syllogism. Thus, the hetu is the property known to occur
in the subject (paksa), known to occur in the instances
similar to the subject (sapaksa) and known to be absent
from the instances dissimilar to the subject (asapaksa).
It is important to realize that Dignaga introduced in his
logic the possibility of quantification. The hetu can oc-
cur in the whole of the sapaksa or in a part of the
sapaksa, and similarly, can be absent from the whole of
the asapaksa or from part of the asapaksa. The relevant
verse is PS III.9
sapakse sann asan dvedha paksadharmah punas tridha /
pratyekam asapakse ca sadasaddvividhatvatah //
(source Randle Frag, p. 29)
[The reason as] a property of the subject is — regarding
the sapaksa [, class of instances similar to the subject,]
-- present [in it], not present, [or] in two ways [ie
present in a part of the sapaksa, not present in another
part] and in so far as [the property of the subject] is
present in, absent from, [or] in two ways [present] in [a
part of the class of] instances dissimilar from the sub-
ject [, the asapaksa]; in each one [of the possibilities,
as regards the sapaksa, we can generate] in three ways
75
[three combinations]f.
The extreme brevity of the original necessitates this
rather lengthy paraphrasing translation in order to lay
bare the intentions of this verse. First of all, it must
be borne in mind that sapaksa refers to the class of
objects in which it is seen that the probans is in-
variably concomitant with the inferendum, and that the
asapaksa is the class of objects in which it is observed
that the probans never occurs with the inferendum. The
term sapaksa also occurred in Vadavidhana (fragm 7, 7a),
but the term asapaksa occurred there as vipaksa. There
are, according to Dignaga, three possible relations of the
probans with the sapaksa or the asapaksa. The probans is
present with the inferendum in the whole of the sapaksa,
is fully absent from it, or is present in some part of the
sapaksa, which is saying as much as being absent from some
part of the sapaksa. The same three possibilities obtain
for the asapaksa; the probans accompanied by the inferen-
dum fully occurs in all of the asapaksa, does not at all
occur in the asapaksa or occurs in some part of the
asapaksa. The probans occurring in the subject and the
three types of occurrence, of the probans in the sapaksa
and asapaksa respectively make in all nine different
combinations. These nine consitute the famous wheel of
nine reasons. ' The first condition is, without exception,
that the probans is a property of the subject, while it is
with the second and third condition that the nine combina-
tions can be made. As regards the sapaksa the three pos-
sibilities are:
a. present in the whole of the sapaksa
b. absent from the whole of the sapaksa
c. present in some part of the sapaksa
With the asapaksa we have the same three possibilities:
d. present in the whole of the asapaksa
e. absent from the whole of the asapaksa
f. present in some part of the asapaksa
The nine pairs we can make are as follows:
1. a and d 4. b and d 7. c and d
2. a and e 5. b and e 8. c and e
76
3. a and f 6. b and f 9. c and f
Only combinations 2 and 8 yield a valid inference, that
is, when the probans together with the inferendum is
present in the whole of the sapaksa and fully absent from
t n e asapaksa, or when the probans is present in some part
of the sapaksa and fully absent from the asapaksa. In
Frauwallner 1959f appendix we find in the Tibetan transla-
tion of Dignaga's Hetucakra a worked out set of 9 syl-
logisms with these nine hetus in Tibetan. Combination 2
reads;
sgra ni mi rtag ste /
byas pa^i phyir /
bum pa bzin dan /
nam mkha^ bzin no /
mthun pa^i phyogs thams cad la yod /
mi mthun pa^i phyogs la med /
rtags yan dag go /
'Sound is non-eternal (thesis)
because it is produced (reason, probans)
like a pot (sapaksa)
[sound is not unproduced] like space (asapaksa)
[the probans] is present in the whole of the sapaksa
not present in the asapaksa
the reason (rtags, hetu) is correct1.
The fact of being produced occurs in the whole of the
sapaksa, for pots and the like are all seen to be produced
and non-eternal. The fact of being produced does not occur
at all in the asapaksa, for space and the like are well-
known to be unproduced as well as eternal. In the sapaksa
we observe the invariable concomitance of the probans with
the inferendum and the extension of the inferendum and the
probans. In this syllogism, the inferendum has the same
extension as the probans, for everything that is non-
eternal is produced and everything that is produced is
non-eternal. When the sapaksa would show that the exten-
sion of the inferendum is greater than the probans, the
syllogism is also valid, because wherever the probans oc-
curs there the inferendum occurs too. This can be ex-
emplified by the smoke-fire syllogism, where we see in the
77
sapaksa that the probans always occurs with the inferendum
fire, and yet the inferendum itself can occur without the
probans, smoke. The example of combination 8 is as
follows:
sgra ni mi rtag ste /
rtsol ba las byun ba"i phyir /
bum pa dan glog bzin dan /
nam mkha^ bzin no /
mthun pa^i phyogs la gnis ka /
mi mthun pa^i phyogs la med /
rtags yan dag go /
'Sound is non-eternal
because it is caused by [some intentional]
effort like a pot and a flash of lightning
[sound is not] like space [which is not caused by inten-
tional effort]
[the probans] is in two ways [present] in the sapaksa
not present in the asapaksa
the reason is correct.1
Although the fact of being caused by some intentional
effort (prayatnotthita) is invariably concomitant with
non-eternity, it does not occur in the whole of the
sapaksa. For the sapaksa actually contains the class of
objects that are produced (krtaka), but are not all neces-
sarily produced by intentional effort. Pots are made with
intentional effort, but lightning is not, though it is
produced by some cause. The fact of being caused by inten-
tional effort has a smaller extension than being-produced.
The latter fully contains the former, so that we could say
that intentional effort is invariably concomitant with
production. And since production is invariably concomitant
with non-eternityf intentional effort is also invariably
concomitant with non-eternity. The third condition is com-
pletely met, the probans does not occur at all in the
asapaksa. Space — being not non-eternal — is not
produced and not caused by intentional effort. Thus, the
probans being present in the subject, in a part of the
sapaksa and absent from the whole of the asapaksa, yields
a valid syllogism.
78
In all the remaining 7 cases, the reason generates
incorrect syllogisms. The reason is present in the
subject, but the other two criteria are either not, or
only partly, met. There is one case, namely 5, where the
probans does not occur in the sapaksa nor in the asapaksa.
Such a reason is called asadharana, uncommon reason, a
reason peculiar to the subject alone, a reason not sup-
ported by any example. In the cases 4 and 6, the probans
is not present in the sapaksa but present in the whole or
in a part of the asapaksa. The reason is then called
viruddha, contradictory, proving the contrary of what one
wished to prove. In cases 1, 3, 7 and 9, where the probans
is present in the whole or in a part of the sapaksa as
well as in the whole or in a part of the asapaksa, the
reason is called anaikantika, inconclusive, because the
reason does not conclusively establish the presence of the
inferendum in the subject. In all these cases, the logical
error arises from the circumstance that the probans is not
seen to be invariably concomitant with the inferendum in
the sapaksa and not seen to never to occur with the in-
ferendum in the asapaksa.
When in the course of the syllogism the sapaksa and
the asapaksa are mentioned, they are regarded as
drstantas, generally-accepted facts. The function of the
drstanta is explained in PS IV. Dignaga opens that chapter
with the following statement.
trirupo hetur ity uktam paksadharme ca samsthitah /
rudhe rupadvayam sesam drstantena pradarsyate //
(PS IV.1 occurs in Vip, p. 88.27-28)
'It is explained [what] the triple reason is [in PS II-
III] and [that the reason] is established as regards [its
presence in the subject as] a known (rudha) property of
the subject [the first criterion]. [As for] the other two
criteria, they are pointed out by means of a drstanta1 .
The other two critera are, of course, the condition that
the probans must always be found with the inferendum in
the sapaksa and that wherever the inferendum is absent
the probans is also absent, a fact that is seen in the
asapaksa. We disclose the positive invariable concomitance
79
of the probans with the inferendum by means of the
sadharmya-drstanta which really is the sapaksa. We dis-
close the negative invariable concomitance with the help
of the vaidharmya-drstanta which is the asapaksa. Dignaga
continues:
sadhyanuqamo hetuh sadhyabhave ca nastita /
chos mthun cig sos dan bcas gnis //
dpe la rab tu bstan par bya //
(PS IV.2a-b source Tucci 1930, p. 21; 2cd Kitagawa 1965f
text V p. 518.3-4) 'The reason [is the property that
always] follows what is to be proved (sadhya) and [the
reason] is not present when what is to be proved is
absent. The fact of possessing the same property (chos
mthun, sadharmya), which [fact] is related to the other
two [criteria of a valid reason], must be shown in a
drstanta (dpe)'. The fact of possessing the same property,
sadharmya, relates to the sapaksa in which we see that the
inferendum is connected with the probans, and thus the
sapaksa possesses the same proving property as the sub-
ject, which is here called sadhya in order to emphasize
that what is to be proved by the hetu is the residence of
the inferendum in the subject (cf PS III.10 quoted above).
In the commentary on this verse Dignaga says,
chos mthun pa ni re zig sgra mi rtag ste byas pa"i
phyir zes bya ba lta bu"oy1 In the first place, sadharmya (chos mthun pa), the fact
of possessing the same property [as the subject functions]
thus: sound is non-eternal because it is produced
(Kitagawa 1965, text V p. 518, 6-7)'. This phrase ex-
emplifies the reason in a syllogism (again the famous syl-
logism on the non-eternity of sound). The next step is the
enunciation of the drstanta,
gal te gan rtsol ba las byun ba de mi rtag par bum pa la
sogs pa la mthon ba de bzin du chos mi mthun pa la yan
rtag pa rtsol ba las ma byun bar nam mkha^ la mthon ba^i
phyir ^dir van sgrub byed kyi sgo nas bsgrub bya sgrub par
brjod par bya^o.
'[And further] if it is seen in [the sapaksa, the
similar instances like] pots etc [used as drstanta]
80
what is caused by [intentional] effort is non-eternal,
[then] in like manner regarding the fact of not possess-
ing the same property (chos mi mthun pay vaidharmya) it
is seen in [the asapaksa, the dissimilar instances like]
space that what jjs eternal is not caused by [intentional]
effort. Therefore, in the [vaidharmya-drstanta] too one
has to state that the proof of the sadhya is [being done]
by the probans (sadhana, sgrub byed) ( Kitagawa 1965,
text V p. 513.7—11)'. As we will remember, the use of a
positive example and a contrary example in the syllogism
was taught in NS 1.1.36-37 and the NBh thereon. According
to those texts, the reason is established by a sadharmya-
udaharana and a vaidharmya-udaharana. Dignaga has
retained this practice also in the PS, for here the
reason is based on a sadharmya-drstanta, showing the
positive invariable concomitance of the probans with the
inferendum as well as a vaidharmya-drstanta, showing the
negative invariable concomitance which evidences the ab-
sence of the probans at the absence of the probandum.
2.11 Reliable Statement
The only two means of valid cognition, according to
Dignaga (and probably Vasubandhu), are perception and
inference. Thus, Dignaga denies all the other pramanas
the status of real means of valid cognition. The number
of two valid pramanas was -- as we have seen — based on
the idea that there exist only two sorts of objects, viz
the particular (svalaksana) and the universal (samanya-
laksana). The particular is cognized by perception, the
universal by inference (cf PS 1.2). This seems to leave
no room for reliable statement, apart from being con-
sidered a kind of inference, as a source of knowledge
about reality. The most perfect form of reliable state-
ment would be the syllogism, the inference put in words.
But even if a reliable statement does not assume the form
of a correct syllogism, Dignaga wishes to include it in
inference. For in PS II.5a-b he mentions reliable state-
81
ment which he calls aptavada, 'discourse, statement1 of
an fexpertf.
aptavadavisamvadasamanyad anumanata
(source PVSV Gn p. 108.1, 109.5)
'The statement by an expert is [to be regarded as] in-
ference in so far as there is similarity in trustworthi-
ness [between real inference and reliable statement]1.
The cognition which results from both reliable statement
and inference seizes upon the universal, and in so far as
the cognition derived from reliable statement is valid,
it is qualitatively similar to inferential cognition --
at least this seems to be the drift of Dignagafs argument
in PS II.5a-b. The commentary does not say much more, al-
though we discern a tendency to base the validity of
aptavada on direct perception.
yid ches pa^i tshig nid bzun nas kyan mi bslu bar
mtshuns pa^i phyir de van rjes su dpag pa nid du brjod
do / de skad du yan / min gi las rnams kyi don du mnon
sum son ba^i phyir ro zes^byun no,
'And further, after we perceived [ie heard] only a state-
ment of an expert, we call also this [kind of statement]
inference [ ie of inferential nature] on account of the
similarity in trustworthiness [between inference and
reliable statement]. And thus [by way of trustworthiness,
a reliable statement] is produced because the activities
of giving names [to things] is [inevitably] preceded by
direct perception (mnon sum, pratyaksa) (Kitagawa 1965,
text V, p. 455.1-4)'. Dignaga is actually saying here
that the trustworthiness of statements is based on the
fact that the objects they refer to were really perceived
by the speaker. Thus, direct perception (pratyaksa) is
the source of valid inferences as well as correct state-
ments that do not assume the form of an inference. And
since the validity of inferences (and thus of reliable
statements) can be verified by the very fact that correct
inferences are drawn in accordance with the rules of
logic, Dignaga does not have to further discuss the
reliability of the speaker, as Paksilasvamin did in NBh
223.5-7. The correct inference depends for its validity
82
on the impersonal laws of valid deduction, not on the
moral qualities of its speaker.
83
NOTES
1. Thi£ article is reprinted in Frauwallner, Kleine
Schriften, Wiesbaden 1982, p. 716-758. The article itself
is numbered there p. 2-44, whereas in the journal in which
it was originally published it occupies p. 104-146.
2. In fact, the notion of invariable concomitance (avina-
bhava, nantariyakatva) as the basis of valid inference
does occur in the NBh, although not in the description of
the inference and the syllogism, where we would expect it.
In NS 2.2.2 various other pramanas are subsumed under
reliable statement or inference. The pramana called
sambhava, 'the fact [that two things] are together1 is
subsumed under inference. Paksilasvamin comments on this
with the following remark,
avinabhavavrttya ca sambaddhayoh samudayasamudayinoh
samudayenetarasya grahanam sambhavah, tad apy anumanam eva
(NBh, p. 229.5-6), 'The fact [that two things] are
together [, as a separate pramana] consists in grasping
the second [object, ie the part] by [grasping] the whole
by [way of] the presence of the never occurring without
(avinabhava) [ie always occurring together, hence invari-
able concomitance] in a whole and its part that are both
connected [with each other] ; also this [kind of grasping]
is only inference [and not a separate pramana ] ' .
Paksilasvamin says here that if a part and the whole of
which it is a part are so connected that the whole never
occurs without its part, that is to say that the two are
invariably concomitant, it is possible to infer the pre-
sence of the part through the presence and observation of
the whole. In his commentary on NS 2.2.1 Paksilasvamin ex-
plains sambhava thus,
sambhavo namavinabhavino 'rthasya sattagrahanad anyasya
sattaqrahanam. yatha dronasya sattaqrahanad adhakasya
sattaqrahanam adhakasya sattagrahanat prasthasyeti
(NBh, p. 228.2-4), 'What is called sambhava is grasping
[ie perceiving] the presence (satta) of another [ie a
second] [thing] through grasping the presence of a thing
84
that is never occurring without [the second thing, ie that
is invariably concomitant with the second thing], as
grasping the presence of [the weight of] an adhaka [1/4 of
a drona] through grasping the presence of [the weight of]
a drona [4 adhakas with which 1 adhaka is invariably
concomitant, and grasping the presence] of [the weight of]a prastha [1/64 adhaka] through grasping the presence of
an adhaka [64 prasthas, with which 1 prastha is invariably
concomitant]1. Paksilasvamin says here that a smaller
weight is always concomitant with, always included in, a
greater weight. For instance, 100 grams of sugar is in-
variably concomitant with, and included in, 500 grams of
sugar. Vasubandhu, in all probability, used this doctrine
of avinabhava between two things to describe in Vadavidhi
the basis of valid inference in general, not only of
saifibhava as a special type of inference as Paksilasvamin
does.
3. NBh, p. 84.1-2, sadhyam ca dvividham dharmivisisto va
dharmah sabdasyanityatvam dharmavisisto va dharmi anityah
sabda iti,
'That which is to be proved (sadhya) is twofold: either it
is a property characterized by a property-bearer, [as the
property] non-eternity of [a property-bearer] sound, or it
is a property-bearer characterized by a property, [as the
property-bearer] sound is non-eternal [ie characterized by
the property non-eternity]'. In the first example, the
property that is to be proved to occur in the property-
bearer is called sadhya; in the second, the property-
bearer of which it must be proved that a certain property
occurs in it is called sadhya.
4. Cf note 2 about this term.
5. Cf the description of NS- and NBh-logic in the previous
chapter and note 22.
6. At least in this description of the syllogism. But he did
know about invariable concomitance as the basis of some
85
special type of inference, sambhava, cf note 2. It is the
achievement of Vasuandhu to have made avinabhava the basis
of correct inference in general and of syllogistic
reasoning.
The MImamsa author Kumarila and the Vaisesika author
Prasastapada, for instance, borrowed a great deal of their
logic from Dignaga. The Nyaya author Uddyotakara heavily
attacked Dignaga and yet made good use of Dignagafs logic,
cf Oberhammer 1964b, p. 136-137 and Steinkellner 1961, p.
152-153.
In an appendix without page numbers, Hattori 1968 gives
the surviving Sanskrit fragments of PS 1.1-11 with the
missing parts supplied from Tibetan. Thus we have a com-
plete running text partly in Sanskrit partly in Tibetan.
Whenever I quote PS I it is from this mixed text of Hat-
tori 1968.
It is strange to find in Hattori 1968, p. 76 that the
first statement of the NBh (NBh 1.5) should have an af-
finity with this assertion of Dignaga. NBh 1.5 says,
pramanato 'rthapratipattau pravrttisamarthyad arthavat
pramanam,fA means of valid cognition possesses the object, because
of the appropriateness of [human] activity [with regard to
the object] when the object is cognized through a means of
valid cognition1. This passage is not similar to Dignaga*s
statement, for NBh 1.5 tries to explain the practical
utility of the pramanas; but my quotation (NBh 2.1) and
Dignaga's statement are indeed affirming the same
principle, namely, that true knowledge of an object
depends on and is produced by pramanas.
In the Nyayamukha, Dignaga says, '... there are only two
pramanas, I mean: inference and direct perception ...
since [the other pramanas admitted by different schools]
such as tradition (sabda), analogy (upamana) etc. are in-
cluded in these two. Thus there are only two pramanas, by
86
which we can apprehend the thing in itself (svalaksana)
and its generality (samanyalaksana)', Tucci 1930, p. 50.
11 . Whenever I am referring to the commentary on the PS, I am
referring to Dignaga*s own commentary on the verses.
12. According to Frauwallner 1961, p. 131-132 this is the same
Vasubandhu who wrote the Vadavidhana, Vadavidhi,
Vimsatika and Trimsika. That Dignaga was well-acquainted
with the Abhidharmakosa is an established fact, because he
wrote an abridgment of that work called Abhidharmakosa-
Marmapradipa which exists in a Tibetan translation in TTP
vol. 118, Thu 144a.8-286b.1 (cf Hattori 1968, p. 3, 8).
13. In PS, Dignaga does not say more on yoga and the ex-
perience of yogins, yogipratyaksa than PS I.6b-c. For fur-
ther details on his conception of yoga we could turn to
his Yogavatara (published in Frauwallner 1959, p. 144-
145). In this little tract Dignaga says that the yogin,
after having heard the elevated doctrine (v. 1), having
shaken off all kinds of concepts (v. 2), having regarded
everything as illusory and having torn to pieces that
which binds him to the body (v. 4), sees his own mind or
pure consciousness (citta) and the highest reality:
svakaramatrasesam pasyati cittam svam adyanutpannam /
yenapi pasyatldam tad api tathaivavalokayati // 5 //
so 'nupalambho 'cintya tathatokta bhutakotis ca // 6ab //f[The yogin] sees that his own consciousness [or mind,
citta], having as its remainder (sesa) [after all concepts
etc are shaken off] nothing but (matra) the form (akara)
of itself (sva), has never originated from a beginning
[ ie the pure consciousness is beginningless]. This
[consciousness] too, through which [the yogin] sees the
[beginningless consciousness], he beholds in the same way
[ namely, as beginningless] 5.
This non-perception [ie beginningless consciousness only
beholding itself as such] is called [in Prajna Paramita
Sutras] the inconceivable [or unthinkable, acintya]
reality (tathata) and the highest limit of being
87
(bhutakoti) 6ab*. This should give us some idea of what
Dignaga has in mind when he speaks of perception / ex-
perience of yogins, yogipratyaksa.
13a I have interpreted the locative mrgatrsnikadisu in the
first sentence as referring to bhrantijnana. However, the
locative could also refer to kalpana. Then the sentence
would state, fAs regards this [list], erroneous cognition
has the [false] appearance of perception because
[erroneous cognition] has become active through the con-
ceptualization of water etc with regard to [things that
are not water etc but in fact] mirages etc1. If this
second interpretation of the locative is applied to the
parallel phrase samvrtisatsu in the next sentence, ie if
we regard it as referring to adhyaropa and/or kalpana in-
stead of referring to samvrtisaj j nana, then we get the
rather surprising and indeed improbable notion that con-
ceptualization — which in fact produces samvrtisat out of
things that exist in the ultimate sense, paramarthasat —
acts on its own product (samvrtisat) in order to produce
precisely that, namely samvrtisat. The sentence would ab-
surdly, maintain that 'cognition of what exists [only] con-
ventionally has the [false] appearance of perception be-
cause [this cognition comes into existence through]
[mental] superimposition [on a real thing] of a thing that
is different [from the real thing]; [ie samvrtisaiinana]
has become active through the conceptualization of the
form of the [conventionally existing things] with regard
to the [things] that exist [only] conventionally1. The in-
terpretation I have adopted — ie samvrtisatsu as refer-
ring to the immediately preceding samvrtisaj1nana — does
not yield a very elegant statement either, but it is still
preferable to the one just mentioned. It could be said
that the Sanskrit original, samvrtisajjnanam samvrtisatsv,
is clumsily formulated (it must be the correct reading as
it is supported by the Tibetan translations, cf Hattori
1968, p. 180.36-37, p. 181.35-36).
I have interpreted the ablative tadrupakalpanapra-
vrttatvat (a) as the reason for regarding samvrtisai1nana
88
as false perception and (b) as an explanatory paraphrase
in the same case of arthantaradhyaropat. It seems per-
fectly evident that adhyafopa has the same function as
kalpana, namely to mentally superimpose or conceptualize
a thing that exists only conventionally on a thing that
exists in the ultimate sense, eg to superimpose "pot" on a
heap of atoms or to form the concept "pot" about a heap of
atoms. Dignaga may have used the term adhyaropa to indi-
cate wherein samvrtisaj i nana originates, and may have
paraphrased adhyaropa by kalpana in order to make quite
clear that samvrtisajjnana is no true perception because
it is not free from conceptualization (kalpana). The only
other interpretation of the two ablatives would be that
arthantaradhyaropat expresses the cause of kalpana. The
conceptualization of the form of samvrtisat would then be
brought forth by a mental or conceptual superimposition of
a different thing on the real thing. This would amount to
the conceptualization of a mental superimposition (ie
again a conceptualization) of a real thing as something
that exists only conventionally. As a double concep-
tualization of a thing seems very unlikely in this con-
text, we could safely stick to the first interpretation.
14. Most of the Sanskrit fragments of PS II and III I owe to
the kindness of Prof. Steinkellner in Vienna who gave me a
copy of his unpublished collection of fragments of the PS.
Some of these fragments are quite well-known and occur in
Kitagawa 1965 or Randle Frag. After the Sanskrit fragment
I will cite the source, while the Tibetan text is taken
from the edition in Kitagawa 1965.
15. In Tillemans 1984 it is argued that the pararthanumana as
taught in the PS and PV should not be called syllogism. On
p. 94 Tillemans states, 'Une inference - pour - autrui
PS-PV n'est pas un "syllogisme" bouddhique, puisque ses
criteres de validite ne ressemblent pas a ceux d'un syl-
logisme dans une loqigue formelle", and on p. 87 he says
'... les differences fondamentals qui separent une
inference - pour - autrui et un syllogisme aristotelicien
89
... la validite d'un syllogisme ne depend que de sa forme
logique, tandis que la cognition ... joue un role majeur
dans la validite selon la logique bouddhique. ... un syl-
logisme ... doit avoir une conclusion, alors que c'est un
point logicophilosophique fondamental de la version PS -
PV que lfexpression de la these (paksavacana) ne figure
pas dans l'enonce d1 une inference - pour - autrui.• Now
his first objection to calling pararthanumana a syllogism
seems to be that a pararthanumana, an inference for the
sake of others, does not constitute a system of formal
logic in the modern sense of the word, which is, of
course, true. It would be misleading and incorrect to call
Dignaga and Dharmakirti's logic a formal logic, for its
main purpose is to arrive at correct knowledge of reality.
Thus their logic is a pramana, a means of valid cognition.
However, I am not so sure that Aristotelian logic, ie the
Aristotelian syllogism could simply be regarded as part of
a system of formal logic in which the correctness, the
validity of the syllogism merely depends on its logical
form, as Tillemans maintains. Moreover, Tillemans does not
make clear whether he means the syllogism as taught by
Aristotle himself or the medieval developments based on
Aristotle. In any case, it seems that Aristotle meant his
logic to be an instrument to gain correct knowledge, cf
the following remark in Kneale p. 67, 'Aristotle's account
of general propositions prepares the way for his doctrine
of the syllogism. At the beginning of the Prior Analytics
he defined a syllogism as a discourse in which from cer-
tain propositions that are laid down something other than
what is stated follows of necessity. This formula is wide
enough to cover almost any argument in which a conclusion
is inferred from two or more premisses1. What is important
here is that from general true propositions another propo-
sition necessarily follows. If this general notion is kept
in mind, we are able to see the correspondence of this
with the pararthanumana, for in the latter the drstanta
functions as a proposition enunciating a general fact
while the hetu enunciates a particular fact. The thing
that was announced for proof follows from both proposi-
90
tions. Tillemans is right when, with regard to Buddhist
logic, he says that cognition (ie direct perception) plays
a major role in the validity of the statements. But I
think he is not quite right in saying that the absence of
a thesis (on which his interpretation of the pararthanu-
mana as not being a syllogism seems to rest) is a funda-
mental logical and philosophical characteristic of the
pararthanumana, As we will see in my description of the
pararthanumana, Dignaga discusses some sort of enunciation
of what is to be proved (sadhyanirdesa) in PS III.1cd-2,
although he makes it clear that this is not a separate
step in syllogistic reasoning as his predecessors thought.
I would say that, for convenience's sake, we could still
use the word syllogism for pararthanumana, provided we
keep in mind that the pararthanumana does not have the
actual form of the Aristotelian syllogistic, nor is it a
kind of formal logic in the modern sense.
16. A class can be regarded as an indefinite number of in-
dividual occurrences or, in other words, a collection of
objects, itself considered as a single abstract object, a
set. In our case, this set or class consists of all the
occurrences of the invariable concomitance of the probans
with the inferendum and the negative invariable concomi-
tance of the probans with objects dissimilar to the
subject. If the notion of sapaksa is taken to denote a
class (as in Frauwallner 1959, p. 85 and more explicitly
in Oberhammer 1964, p. 136-138) we can refer to sapaksa. in
the singular. If the sapaksa is taken to denote a number
of similar instances (as in Randle Frag, p. 30), we can
refer to it in the plural. Thus, saying that the probans
is present in all the sapaksas or in the whole of the
sapaksa, really amounts to the same proposition and,
similarly, to saying that the probans occurs in some
sapaksas or in a part of the sapaksa. But one should be
careful in using terms like fallf and 'some1 because they
suggest real logical quantifiers in the original, whereas
Dignaga never used any quantifiers in the Aristotelian
sense of the word. Dignaga himself, at all events, refers
91
to sapaksa and asapaksa in the singular.
17. Cf Vidyabhushana, A History of Indian Logic, reprint Delhi
1971, p. 283-285; Randle Frag, p. 31-33; Tucci 1930, p.
26-30; Frauwallner 1959, p. 85-87; Hayes 1980, p. 230-237.
According to Frauwallner 1959, the Hetucakradamaru which
expounds the nine reasons is one of Dignaga's earliest
writings on logic, cf op cit p. 89.
18. It is interesting to compare Dignaga's syllogistic with
the illustrative syllogism on the non-eternity of sound
offered by Paksilasvamin in his commentary on NS 1.1.33-
39. In spite of the absence of the upanaya and nigamana,
Dignaga's syllogism has the same structure as Paksila-
svamin 's. One could say that Dignaga submits a highly
improved version of Paksilasvamin1s syllogism.
93
III Dharmaklrti's Logic
In spite of his widespread renown, Dignaga is greatly
overshadowed by his philosophical successor Dharmaklrti/
who lived later, ca 600-660 AD according to Frauwallner
1961, p. 137-139, but according to Lindtner 1980, p. 29,
ca 530-600 AD. Dharmaklrti was the one who gave definite
shape to Buddhist epistemology and logic by his various
works, of which PV I with PVSV and PV II-IV are the
earliest. PV II-IV were intended to be an exhaustive com-
mentary on the PS, a commentary in which Dharmaklrti also
allowed himself to elaborate his own epistemological and
logical theories. PV II comments on the first verse of PS;
PV III refers to PS I, while PV IV refers to PS III, PV I
with PVSV are shown by Frauwallner to be the oldest part
of PV, coming before PV II-IV. PV I together with PVSV is
an independent treatise on topics related to PS -- parti-
cularly logic and semantics — without being a direct com-
mentary (Frauwallner 1954, p. 146, 148, 152). Not only is
PV I a separate work, it is, moreover, Dharmaklrti's first
treatise (Frauwallner 1954, p. 152-153), and PV II-IV can
only be fully understood in the light of PV I and, of
course, of Dignaga's philosophy.
3.1 The utility of inference
Before finally examining PV II itself, it is necessary to
show how Dharmaklrti in PV I with PVSV fundamentally
revised and improved Dignaga's theory of inference. The
first prose sentence of the PVSV specifies the utility of
drawing valid inferences (anumana).
arthanarthavivecanasyanumanasrayatvat tadvipratipattes
tadvyavasthapanayaha
(PVSV, p. 1.8-9). 'In so far as the distinguishing between
what is meaningful (artha) and what is not meaningful
(anartha) [in epistemology and logic] has [correct] in-
ference as [its] basis, [and] because there is perplexity
[ie a variety of conflicting opinions] about it [ ie about
correct inference], [therefore] he [ ie Dharmaklrti] says
94
[the following verse, PV I.1] in order to [clearly] deter-
mine it [is inference]1. One of the pramanas that is used
to acquire valid knowledge about reality is inference,
which primarily means internal inference, svarthanumana,
but also refers to the syllogism, pararthanumana, as the
latter is, in Dharmaklrtifs opinion, nothing but an ex-
plicitly formulated inference. The importance of infe-
rence as the only second pramana along with pratyaksa com-
pels Dharmakirti to dissipate the conflicting and con-
tradictory opinions on inference by clearly describing a
correct process of drawing inferences and the circum-
stances in which an inference can be valid. What is impor-
tant to Dharmaklrti is that an inference be valid as a
pramana, that it may give true knowledge about reality;
for it is correct knowledge that enables one to make the
right choices between what is useful and what is not.
3.2 A compendious definition of the probans
In the verse that now follows, Dharmakirti defines what
was called by Dignaga the inferens, linga in the internal
inference and reason, hetu, in the syllogism.
paksadharmas tadamsena vyapto hetus tridhaiva sah /
avinabhavaniyamad dhetvabhasas tato "pare //
(PV 1.1)3
'The reason is a property of the subject pervaded by a
part [ie another property] of the [subject]. The [reason
as valid probans] is only of three kinds owing to the
restriction of the positive invariable concomitance [of
the reason with the probandum]. [Reasons] other than this
[one] are fallacious reasons'. This programmatic verse
contains various propositions:in the words of Sakyamati's
Tika on PV I and PVSV,
tshigs su bead pa ^di las [la] ni rtags kyi mtshan nid
dan / grans nes pa dan /nes pa^i rgyu dan mi mthun pa^i
phyogs las ldog pa bsad de /
(PVT, Je, p. 7.6-7, cf PVSVT, p. 8.24-25), 'in this verse
are expressed; a definition of a [correct] reason, a
restriction of the number [of correct reasons, ie only
95
three], the cause of this restriction [ie the positive in-
variable concomitance of the probans with the probandum],
and the absence [of the correct probans] from [all in-
stances that are] contrary [to the definition of the cor-
rect probans]1. If the reason (hetu) conforms to this
definition, it is a correct reason; if it does not, it is
a false reason (hetvabhasa). Sakyamati explains his last
remark as follows,
gtan tshigs ltar snan de las gzan zes bya ba ni mi
mthun pa^i phyogs las ldog pa ste / gtan tshigs gsum po
de las gzan pa ni gtan tshigs ltar snan ba dag yin no
(PVT, Je, p. 7b.1, cf PVSVT, p. 8.28-29), f "Other [kinds
of reasons] than this [one] are fallacious ^reasons"
[this means:] the absence [of the correct reason] from [all
instances that are] contrary [to the definitions]; [ie
reasons] other than these three [valid] reasons are falla-
cious reasons1. In the above verse, the reason is defined
as a property of the subject, which means that it is a
known property of the subject; this is the first of the
three criteria of a valid inferens (linga) and reason
(hetu) as defined by Dignaga in PS II.5c-d and PS III.9.
Not only must the probans be a known property of the sub-
ject, it must also be invariably concomitant with, and be
pervaded by, the probandum, another hitherto unknown pro-
perty of the subject. This other property 'pervades1 the
reason, which is a different way of saying that its exten-
sion is greater than, or as great as, that of the invari-
ably concomitant reason. What is to be established by the
valid reason is, of course, the presence of the pervading
property in the subject. The probans, in this verse, is
implied to be the threefold probans. The first criterion
is already mentioned; it is a property of the subject. The
second criterion demands that the probans must be known to
always occur with the probandum in the sapaksa (PS II.5c-
d, PS III.9), and the third criterion that the probans in
the asapaksa must be known never to occur when the proban-
dum is not present (PS II.5d, PS IV.2b). All this has al-
ready been discussed in the chapter on Dignaga. According
to Dharmakirti, the second and third criteria are met
96
through the principle of positive invariable concomitance.
If it is known that the reason is fully pervaded by the
probandum, then it is always true that wherever the reason
is found, there the probandum must be found too. And it is
always true that wherever the probandum is not found, the
reason is not there either, precisely on the grounds that
the probans, having the smaller extension, is invariably
concomitant with the probandum. Dignaga already has said
as much in PS IV.2a-bf
sadhyanugamo hetuh sadhyabhave ca nastita
(Tucci 1930, p. 21), fThe reason [is the property that
always] follows what is to be proved (sadhya) and [the
reason] is not present when what is to be proved is
absent1. But Dignaga does not draw from this the conclu-
sion that Dharmakirti draws. In the commentary on PV 1.1,
Dharmaklrti interprets the second and third criterion as
follows,
sajatlya eva sattvam iti siddhe^pi vijatlya-vyatireke
sadhyabhave"sattvavacanavat,
[The statement of the second criterion that the reason --
being a known property of the subject — ] is present only
in the [class of objects] similar [to the subject] is com-
parable to the statement [of the third criterion, ie] of
the absence [of the reason] at the absence of the proban-
dum [when this statement is made] with regard to the nega-
tive invariable concomitance [of the probans with the
probandum] in [all classes of objects] dissimilar [to the
subject], although [this third criterion] is [already] es-
tablished [with the second criterion, namely, with the
statement that the probans occurs only in the class of ob-
jects similar to the subject] (PVSV, p. 2.5-6) '.
Dharmakirti is saying here that the establishment of the
positive invariable concomitance of the probans with the
probandum and the presence of the probans in the subject
suffice for validly inferring the presence of the proban-
dum in the subject. He implies that the explicit mention-
ing of the third criterion of a valid reason is super-
fluous, for the second criterion, ie the positive invari-
able concomitance of the probans with the probandum in the
97
sapaksa, logically entails the negative invariable con-
comitance of the probans with what is not the probandum in
the asapaksa. This aspect of Dharmaklrti's logic will be
further elaborated in the sequel.
About the pervasion (vyapti) of the probans by the
probandum, Dharmaklrti says,
vyaptir vyapakasya tatra bhava eva, vyapyasya va tatraiva
bhavah,
(PVSV, p. 2.12-13), 'Pervasion [of the probans by the
probandum means]: the pervading [property, the probandum]
is only present there [ie with the probans] or [pervasion
of the probandum by the probans means:] the [property that
is] to be pervaded [ie is pervaded, the probans,] is
present only there [ ie with the probandum]1. In this
passage, the two aspects of pervasion are described. From
the point of view of the probans, pervasion means that the
probandum as pervading (vyapaka) property only occurs with
the probans. From the point of view of the probandum, per-
vasion means that the probans as pervaded (vyapya)
property is found only with the probandum.
The 'only three kinds1 of reasons according to PV 1.1
does not refer to the three criteria of a valid reason,
but to a new theory of Dharmaklrti' s own make. PV 1.2-3
define the three kinds of valid reason. A short explana-
tion of this new threefoldness already appears in the
prose following on PV 1.1.
ta ete karyasvabhavanupalabdhilaksanas trayo hetavah,
yatha^gnir atra dhumat. vrkso "yam simsapatvat.
pradesavisese kvacin na ghata upalabdhilaksana-
praptasyanupalabdheh,
'These three reasons [referring to the term tridha in PV
1.1] are defined as effect, essential property^ and non-
perception; [they function as valid reasons] as [in the
propositions:] there is fire here because there is smoke
[effect as probans], this is a tree [in general] because
it is a simsapa [a particular kind of tree, ie a smaller
essential property as probans], at some particular spot
there is not a pitcher because there is no perception of
[the pitcher which is a thing] that has got characteris-
98
tics for cognition [ie can be cognized] (PVSV, p. 2.14-
17)f. Effect as logical reason is illustrated by the well-
known example of fire and smoke. Smoke being an invariably
concomitant effect of fire, with a smaller extension than
fire (because some fire occurs without smoke), can be used
as a valid reason to infer the presence of fire. The fact
of being-a-simsapa, that is, the class of simsapa-trees,
is invariably concomitant with the greater class of trees
in general. The svabhava of simsapa-ness, the essential
property of being a simsapa- tree is logically contained
in, and pervaded by the essential property of being a tree
in general, and hence simsapa-ness can be used as the
probans to validly infer the presence at a given spot of
tree-ness in general. Non-perception, at a given spot, of
a perceivable thing like a pitcher can be used to validly
infer the absence of it at that spot.
As for the logical necessity that in the sapaksa the
probans is observed to be invariably concomitant with the
probandum, Dharmakirti maintains there are only two pos-
sible cases in which the positive invariable concomitance
(anvaya) of the probans with the probandum can be estab-
lished beyond any doubt. Either the probans shares its es-
sence with the probandum or the probans is an invariably
concomitant effect of the probandum. Therefore Dharmakirti
teaches in the first verse of PV I that '[the reason as
valid probans] is only of three kinds, owing to the
restriction of positive invariable concomitance1. That is,
the three above-mentioned kinds of valid reason are all
generated by invariable concomitance, which can only exist
when the probans possesses the same essence as the proban-
dum, or when the probans is brought into being by the pro-
bandum. In both cases, there is a connexion of probans
with probandum through an essence, svabhava. In the words
of Dharamaklrti,
svabhavapratibandhe hi saty artho *rtha?ti na vyabhicarati,
sa ca tadatmatvat,
'When there exists an [inseparable] connexion through an
essence (svabhava), [one] thing [possessing that essence]
never deviates from [ie is invariably concomitant with the
99
other] thing [possessing the same essence]. And that
[connexion] exists on the grounds that [the object acting
as probans] is [in reality] the essence of the [object ac-
ting as probandum] (PVSV, p. 2.19-21)'. As is explained in
the example of simsapas and trees in general, both classes
of objects possess the same essential property of tree-
ness, and they are connected with the same svabhava;
therefore simsapas are invariably concomitant with trees
in general. About the connexion of an effect (acting as
probans) with an essential property, Dharmaklrti says,
karyasyapi svabhavapratibandhah, tatsvabhavasya
tadutpatter iti,
'Also for an effect there exists an [inseparable] con-
nexion through an essence, because the essence of the
[effect] is produced by the [cause] (PVSV, p. 3.3-4)'. We
might say that Dharmaklrti derives the invariable
concomitance, which is the basic principle in his concep-
tion of valid reasoning, from the actual connexion of the
probans with the probandum through an essence. This es-
sence through which probans and probandum are connected is
either their factual essence or else, the essence of the
probans is produced by a specific complex of causes which
complex constitutes the probandum. In the first case, by
probans and probandum the same essence is meant; in the
second case, the essence of the probans can only spring
from the essence of the probandum.
3.3 Effect as probans
The effect (karya) used as probans is explained in PV
I.2a-c.
karyam svabhavair yavadbhir avinabhavi karane /
tesam (PVSV, p. 3.14)
hetuh,
'With as many essential properties the effect is invari-
ably concomitant [and present] in the cause, for those
[essential properties, the effect] is the logical reason,1.
Only an effect that is invariably concomitant with certain
properties of the cause, can be used to validly infer the
100
presence of those properties, as in the example of smoke
and fire. Since smoke is produced by fire alone and is,
therefore, invariably concomitant with fire, it can be
used as logical reason to infer the presence of fire. But
how can a thing be established as the effect of something?
Dharmakirti answers:
yesam upalambhe tallaksanam anupalabdham yad upalabhyate,
tatraikabhave^pi nopalabhyate, tat tasya karyam,
'A [thing] which, unperceived [elsewhere and] possessing
characteristics for it [ ie for perception], is perceived
when other [things] are perceived [and] which is not per-
ceived in the absence of even one [of those other things],
that [thing] is the effect of that [assemblage of things
acting as cause] (PSVS, p. 22.2-3)' .7 The idea is that an
object is the effect of an assemblage (samagri) of other
objects if the said object is not perceived elsewhere but
together with the assemblage, and if the effect is not
perceived when at least one element in the assemblage is
missing. Therefore, it could be said that the object is
brought forth only when the assemblage of causes (karana-
samagri), ie all required causes, are present and
cooperate in producing the effect. In another prose
passage, Dharmakirti elucidates his thought on the as-
semblage of causes a bit more.
kathaift tarhidanim bhinnat sahakarinah karyotpattir
yatha caksurupader vijnanasya. na vai kimcid ekam
janakam tatsvabhavam. kirn tu samagri janika tatsvabhava.
saivanumlyate. saiva ca samagrl svabhavasthityasrayah
karyasya.
'Now, how then is there the arising of an effect from a
particular cooperating [cause] as [for instance the
arising] of cognition from [particular causes like] the
eye, forms etc? Surely there is not some one single
productive cause that possesses the essence of [bringing
forth the effect], but [, on the contrary ,] it is a pro-
ductive assemblage [of causes] that possesses the essence
of [bringing forth the effect]. And that [assemblage]
alone [is what] is inferred. And that same assemblage is
the substratum on which rests the essence of the effect
101
(PSVS, p. 23.18-21)'.8 One particular effect is always
produced by an assemblage of causes, and thus, the exten-
sion of the effect is always equal to, or smaller than the
extension of the cause, ie the assemblage of causes. This
is truly so because the effect arises only when all con-
stituent causes are present. And thus the effect is also
always invariably concomitant (avinabhavin) with the as-
semblage of causes, for the assemblage is necessarily
present when the effect has arisen.
Dharmaklrti sums up his thought on effect as logical
reason (karyahetu) by referring to the causal relation be-
tween smoke and fire.
agnisvabhavah sakrasya murddha yady agnir eva sah /
athanagnisvabhavo^sau dhumas tatra katham bhavet //
dhumahetusvabhavo hi vahnis tacchaktibhedavan /
adhumahetor dhumasya bhave sa syad ahetukah //
(PV 1.36-37)fIf an anthill possesses the essence of fire [namely to
produce smoke], then it is fire. But if it does not pos-
sess the essence of fire, how could there be smoke near it
[ie arising from it]? For fire [alone] possesses the es-
sence of [being] the cause of smoke, [since fire alone] is
distinguished by the power to [produce] it [ie smoke]. If
the occurrence of smoke from [a thing that is] not the
cause of smoke [were possible], it [ie smoke] would be
without cause [, which is absurd]10 (36-37)'.
3.4 Essential property as probans
Essential property (svabhava) as logical reason is
described in PV I.2c-d.
svabhave bhavo^pi bhavamatranurodhini //
'Also something (bhava) [is a logical reason] with regard
to an essential property that conforms to the mere
[actual] existence [of the first thing]1.
What is said here is, that the essential property with a
smaller extension serves as the logical reason for infer-
ring the presence of a property with a wider extension and
in possession of that same essence. In the commentary,
102
Dharmaklrti adds,
tadatmyam hy arthasya tanmatranurodhiny eva nanyayatte,
'For the fact that an object possesses the essence of the
[other object belonging to the same class] is [possible]
only in [an object] conforming to being solely [ie in the
fullest measure] the [class of similar objects], not in
[an object that is] dependent upon another [object or
class of objects for its being] (PVSV, p. 4.2-3)1.11 The
example of a svabhava with a small extension as probans of
a probandum with a wider extension has already been given
before in PVSV, p. 2.16, vrkso"Vam simsapatvat, 'This is a
tree [in general] because it is a simsapa-tree'. Since the
essence tree-ness is present in both classes, the probans
with the smaller extension is invariably concomitant with
the probandum. It is always the case that wherever there
is a simsapa-tree, there is also tree-ness in general, for
simsapa-ness is materially, and therefore logically, con-
tained in general tree-ness. The invariable concomitance
of the probans with the probandum is the fact that the es-
sential property of general tree-ness pervades all in-
dividual classes of trees.
3.5 Non-perception as probans
The third kind of valid probans consists in negating the
presence of a thing at a particular given place
(pradesavisese kvacid) on the grounds that the thing is
not perceived there. First, non-perception in general is
described.
apravrttih pramananam
anupalabdhih (PVSV, p. 4.6)
apravrttiphala^sati / PV I.3ab
'The inactivity of the means of valid cognition is non-
perception which results in the absence of activity [on
the part of the perceiver] with regard to [a thing] that
is not present [at a particular place]1.
Dharmakirti regards non-perception of a thing at a
particular place as the logical sign of the absence of
that thing. Something that is not perceived through any
103
means of valid cognition is not present. To exist
(somewhere) according to Dharmaklrti, is the same as being
perceived. This epistemological doctrine is quite clearly
expressed in the following passage from the commentary on
this verse;
sajjnanasabdavyavaharapratisedhaphala, upalabdhi-
purvakatvat tesam iti. idam sadasatpratisedhavidhihetvos
tulyam rupam. tatha hi sattvam upalabdhir eva
vastuyogyatalaksana tadasraya va jnanapravrttih
(PVSV, p. 4.8-11), [The inactivity of the means of valid
cognition, ie non-perception] results in the negation of
cognizing, naming and treating a [thing as] present [at a
certain place], because these [ ie cognizing, naming and
treating a thing as present] are preceded by the cognition
[of a thing at a certain place]. This [result] is the same
(tulyam rupam) in both [kinds of] reasons [whether they]
negate [something as] present [or] affirm [something as]
not present. In this way the presence [or existence of a
thing] is [its] perception , being defined (-laksana) as
the capability [on the part] of the thing [to be
perceived]; or [the presence of the thing is] the activity
of knowing [this thing, which activity of knowing] is
based on the [capability of the thing to be perceived]1.
Non-perception is not meant to refer to total absence
of any cognition, but to refer to the absence of a
specific cognition with regard to a particular place. Only
absence of a particular piece of cognition can be used to
infer the absence of certain objects.
asajjnanaphala kacid dhetubhedavyapeksaya //
'A specific [inactivity of the means of valid cognition]
results in knowing that [a thing] is not present [some-
where] [only] in dependance on a particular character of
the cause (PV I.3c-d)'. In the commentary Dharmaklrti
says,
hetur anupalabdhih, bhedo^sya visesanam
upalabdhilaksanapraptasattvam,fThe cause is the non-perception [of a thing]. "Particular
character" is a distinguishing property of this [non-
perception], [namely] the existence that has got marks for
104
perception [ie is capable of being perceived] {PVSV, p.
4.14-15) '• Only the non-perception of a perceivable par-
ticular object is a logical reason to infer the absence of
the thing. And thus, only the non-existence or absence of
particular perceivable things can be proved. Dharmaklrti
maintains that a meaningful negation can only be made of
particular things which are perceivable, for a meaningful
negation consists in communicating the non-perception of
something;
yavan kascit pratisedhah sa sarvo^nupalabdheh,
'Any [meaningful] negation whatsoever is [derived] from
non-perception [of something] (PVSV, p. 5.9)'. Dharmaklrti
says further,
tatha sa dvidha kriyeta kasyacid vidhina nisedhena va#
'and that [meaningful negation] can be made in two ways,
either by affirming some [particular] thing or by denying
[some particular thing] (PVSV, p. 5.9-10)'.
3.6 Four kinds of non-perception
PV 1.4 enumerates four ways in which non-perception by way
of affirmation or denial gives knowledge of the absence of
a thing.
viruddhakaryayoh siddhir asiddhir hetubhavayoh /
drsyatmanor abhavarthanupalabdhis caturvidha //
'The establishment [amounting to an affirmation] of a con-
trary [thing] or an effect of a contrary [thing] and the
non-establishment [amounting to a denial] of a cause or an
essence, the essences of [which two pairs, ie a contrary
thing and the effect of a contrary thing as well as a
cause and an essence] are [normally] perceivable, con-
stitutes the fourfold non-perception whose purpose it is
[to prove] the absence [of a particular thing somewhere]1.
Dharmaklrti enumerates the only four cases of non-
perception. The first two consist in perceiving something
contrary to what was expected or an effect of something
that is contrary to an expected cause. The second two con-
sist in not perceiving a cause or not perceiving an
essence. This general principle of fourfold non-perception
105
as a valid probans seems to be based on the two kinds of
essential relations accepted by Dharmaklrti, ie tadatmya,
the fact that two things have the same essence and karya-
karanabhava, the fact that two things are connected as ef-
fect and cause* The absence of something is either estab-
lished directly through its non-perception or the non-
perception of its cause. The absence may also be estab-
lished indirectly through the perception of something
contrary, or the effect of something contrary. By applying
the positive invariable concomitance of an effect with its
cause, or of a greater essence with a smaller one to these
four non-perceptions, a certain number of specific non-
perceptions as probans can be made. In PVSV, p. 6.1-6
Dharmakirti gives eight possibilities. Later, in PVin ten
examples, and in NB II eleven examples are given (cf
Steinkellner 1979, p. 60 note 178).
viruddhasiddhya yatha na sitasparso^tragneh etena
vyapakaviruddhasiddhir ukta veditavya yatha na
tusarasparso^tragneh,
'[A negation based on non-perception is expressed] by es-
tablishing a contrary [thing], as: there is no sensation
of coldness here, because there is fire. By this [sort of
negation] the establishing of a pervading contrary [thing]
must be understood to have been expressed, as: there is no
sensation of frost here, because there is fire (PVSV, p.
6.1-3)'. In the first example, fire is the contrary of the
sensation of coldness. If there is fire, there cannot be
at the same time the sensation of coldness at the spot
where the fire is. The presence of coldness is negated by
referring to the contrary of coldness, ie fire. According
to Dharmaklrti, it is also legitimate to prove the absence
of an invariably concomitant essence with a smaller exten-
sion than the sensation of coldness by referring to the
presence of the contrary of coldness, fire. The sensation
of frost is an invariably concomitant essence of the sen-
sation of coldness with a smaller extension than coldness.
Thus, if fire by its heat drives away coldness, it can
also drive away an invariably concomitant essence, namely
the sensation of frost. If coldness goes, frost inevitably
106
also goes. Fire is the contrary, viruddha, of the pervad-
ing object (vyapaka), the sensation of frost. The text
continues,
viruddhakaryasiddhya yatha na s I ta spars'o"tra dhumat
(PVSV, p. 6.3-4), '[A negation is expressed] by estab-
lishing the effect of a contrary [thing], as: there is no
sensation of coldness here, because there is smoke1. Here
the presence of coldness is negated by the presence of an
invariably concomitant effect of the contrary object,
fire. Since smoke invariably points to the presence of
fire, it also points to the absence of the contrary of
fire, coldness.
So far, the non-perception and absence of a thing are
established by the perception of the contrary thing.
Subsequently, Dharmaklrti describes the actual non-
perception of a thing, which proves its absence.
hetvasiddhya yatha natra dhumo^nagneh,1[A negation is expressed] by the non-establishment of the
cause, as: there is no smoke here, because there is no
fire (PVSV, p. 6.4) '. When we see no fire at a particular
spot, we will certainly see no invariably concomitant
smoke. If the thing with the greater extension is absent,
then the invariably concomitant thing with the smaller ex-
tension is necessarily also absent, ie its presence is
negated. The commentary continues,
svabhavasiddhya yatha natra dhumo^nupalabdheh,
'[A negation is expressed] by the non-establishment of the
essence, as: there is no smoke here, because there is no
perception [of smoke at this spot] <PVSV, p. 6.5) '.
Dharmaklrti continues,
etena vyapakasvabhavasiddhir ukta yatha natra simsapa
vrksabhavat,fWith this [kind of negation] is expressed the non-
establishment of the pervading essence [ ie the essence of
the class with the wider extension], as: there is no
simsapa here, because there is no tree at all [here]
(PVSV, p. 6.5-6) f. The presence of a class of objects is
negated by referring to the total absence of the class of
objects with the wider extension. When, at a certain spot,
107
there are no trees at all, the spot is also necessarily
devoid of simsapas that possess general tree-ness as their
essence.
Dharmaklrti concludes this explanatory passage by
saying,
sarvatra casyam abhavasadhanyam anupalabdhau drsyatmanam
eva tesam tadviruddhanam ca siddhir asiddhis ca veditavya,
'And it must be understood that in all [these cases of]
non-perception proving the absence [of something], there
is the establishment and the non-establishment of only
those [objects] that possess a perceivable nature and of
[those objects that are] contrary to the [objects that are
to be negated] (PVSV, p. 6.6-7)'. A meaningful negation
can only be made of things whose nature it is to be
perceivable. Non-existence or absence can be predicated
only of perceivable things.
3.7 The drstanta
In the classical syllogism before Dharmaklrti, mention had
to be made of two generally-accepted facts, drstantas
showing the positive invariable concomitance (anvaya) of
the probans with the probandum and the negative invariable
concomitance (vyatireka) of the probans with what is not
the probandum. The second drstantaf showing the vyatireka,
must prove that the probans never occurs with things that
are not the probandum. With Dharmaklrti' s new theory of
positive invariable concomitance, these two drstantas# but
especially the second, have become superfluous. The valid
probans as defined by Dharmaklrti suffices for making the
syllogism valid, for the invariable concomitance of the
probans with the probandum is necessarily present when the
probans is the actual product (karya) of the probandum, or
when the probans shares its essence (svabhava) with the
probandum. With this simplification Dharmaklrti deviates
from, and improves upon, the received logic of Dignaga in
PS II-IV, set forth in the previous chapter. Dharmaklrti
provides a solid and unerroneous material basis for the
positive invariable concomitance, which Dignaga did not
108
provide.
About the necessity of a drstanta in the syllogism,
Dharmaklrti says in PV 1.27,
tadbhavahetubhavau hi drstante tadavedinah /
khyapyete, vidusam vacyo hetur eva hi kevalah //1 For in the generally-accepted fact (drstanta), the
[reason's being] the essence of the [probandum] or [the
reason1s] having the [probandum] as [its] cause, are com-
municated to him who does not know that [ie the invariable
concomitance]. To those who know [the invariable concomi-
tance], only the bare reason needs to be told1. Dharma-
klrti reduces the syllogism to a proposition which con-
tains a thesis and a well-founded reason, the sound basis
of the reason being the known invariable concomitance of
the reason with the object acting as the probandum. If the
hearer of the syllogism is not completely familiar with
the invariable concomitance, the speaker could refer to
the positive drstanta, in all instances of which it is
known that the reason invariably occurs with the pro-
bandum. Dharmakirti demands, however, that this drstanta
also shows why the probans and the probandum are in-
variably concomitant, namely in so far as the probans is
the effect of the probandum or has the same essential
property as the probandum. But if the connexion between
the probans and the probandum is known, it is enough to
mention only the reason, because the valid reason will im-
mediately evoke the probandum that is connected with the
reason.
About the positive and contrary drstantas,
Dharmakirti says in PV 1.28 that either one of them im-
plies the other when it is established that the probans is
the effect or the essence of the probandum. Thus only one
of the two drstantas would have to be mentioned in a syl-
logistic proof.
tenaiva 1natasambandhe dvayor anyataroktitah /
arthapattya dvitlye"pi smrtih samupaiayate //
'Therefore when [the probans] has a known connexion [with
the probandum, either by being the effect of the probandum
or by being the essence of the probandum] — recollection
109
arises by implication''3 [in the hearer of the logical
proof] also as regards the second [drstanta] by the men-
tioning of either one of the two [drstantas, ie the posi-
tive and the contrary drstanta]'.
In the prose after this verse, Dharmakirti explains
the function of the negative invariable concomitance
(vyatireka) and its relation with the positive invariable
concomitance. He uses the well-known inference of fire
from smoke as example.
tasmat karyaiti dhuma ity anvayena viditatatkaryatvasya
dahanabhave dhumo na bhavatlty arthad vyatirekapratipattir
bhavati.
'By implication the cognition of the negative invariable
concomitance [of smoke with what is not fire], ie that in
the absence of fire there is no smoke, arises for him who
knows that [smoke] is the effect of that [ie of fire] by
means of the positive invariable concomitance [of smoke
with fire alone] (PVSV, p. 19.8-9) '. The cognition that
smoke can only be produced by fire, and that on these
grounds smoke is invariably concomitant with fire alone,
implies also the cognition that smoke is negatively in-
variable concomitant with what is not fire. When there is
no fire at a certain spot, there is no smoke, from which
fact conclusively follows that smoke is only produced by,
and only occurs with, fire.
tatha"saty agnau dhumo nastlty ukte"gnir dhurrie bhavaty
avasyam ity arthad anvayapratipattih,
'Thus, when it is said that in the absence of fire there
is no smoke [which statement constitutes the negative
drstanta], there is by implication the cognition of the
positive invariable concomitance that there is necessarily
fire in [case there is] smoke (PVSV, p. 19.10-12)'.
There is a verse which neatly epitomizes
Dharmakirti's theory on the real basis of positive invari-
able concomitance and the use of the two drstantas.
karyakaranabhavad va svabhavad va* niyamakat /
avinabhavaniyamo ^darsanan na na darsanat //
(PV 1.31)
'The restriction of the positive invariable concomitance
110
[of the probans with the probandum is possible] either on
account of a restrictive occurrence of effect and cause
[, restrictive in the sense that the effect is invariably
concomitant with the cause,] or on account of a restric-
tive essence [, restrictive in the sense that the probans
is the essence of the probandum]. [But the restriction of
the positive invariable concomitance is] not [established
beyond doubt merely] on the grounds [that two things] are
seen [together in the sapaksa] and not seen [together in
the aspakasa]'•
Here, it is explicitly stated that the restriction of the
positive invariable concomitance is protected when two
things are related as effect and its cause, or when two
things have the same essence — which makes them belong to
the same class of objects. And from this it follows that
all instances in which two things are seen together and
all instances in which two things are not seen together
cannot prove the invariable concomitance of the two
things. For Dharmaklrti's logic demands that positive in-
variable concomitance, being the basic principle that
makes all reasoning valid, is a real connexion such as ex-
ists between products and their causes and things sharing
the same essence.
111
NOTES
1 . This interpretation (one among several possibilities) is
based on an important passage in the PVT, a passage to
which Prof. Steinkellner in Vienna has drawn my attention.
don ni slob dpon phyogs kyi glan pos mdzad pa"i tshad ma
yi mtshan nid la sogs pa yin te rigs pa dan ldan pa^i
phyir ro // don ma yin pa ni mu stegs kyis byas pa yin te
/ rigs pa dan mi ldan pa*i phyir ro //
(PVTr p. 6b.6-7) (cf Sanskrit fragm),
acaryadignagapranltam pramanalaksanadikam artho yuktatvat.
tlrthikapranitam na yuktatvad anarthas
(PVSVT, p. 6.2-3), fWhat is meaningful is [ie refers to]
[the epistemological science] starting with the defini-
tions of the means of valid cognition as promulgated by
the teacher Dignaga, because [his epistemological
thinking] is correct. What is not meaningful is [ie refers
to] [the epistemological science] as promulgated by the
adherents of other doctrines [ie the non-Buddhist
philosophers] because [their epistemological, thinking] is
not correct.f
2. As will become clear in the seguel of the PVSV, Dharma-
klrti describes inference without differentiating between
internal inference, svarthanumana, and the syllogism,
pararthanumana, since both are based on the same logical
theory, cf also Steinkellner 1967 II, p. 82, 'Die von
Dignaga tlbernommene Unterscheidung von svarthanumana und
pararthanumana wird bei Dharmaklrti beibehalten, doch
fSllt fttr ihn, filr den die Verknflpfung allein das
wesentliche Thema der Logik ist, der Unterschied weit-
gehend weg'.
3. According to Frauwallner 1954, p. 145 the beginning of
this verse must come from a lost work of Dignaga1s, as it
is quoted by Uddyotakara in the Nyayavarttika, NV, p*
131.17, in a slightly different wording:
grahyadharmas tadamsena vyapto hetuh.
112
4. According to Frauwallner, Sakyamati could have lived ca
660-720 AD and was a pupil of Devendrabuddhi, a direct
pupil of Dharmakirti, Frauwallner 1961, p. 145.
5. The same passage is repeated in Hetubindu and is commented
upon in Steinkellner 1967 II, p. 90, "Eine Umfassung liegt
seitens dessen, was umfasst ... vor, wenn die umfassende
Beschaffenheit da, das heisst an dem Beschaf fenheitstra1-
ger, an welchem auch die umfasste Beschaffenheit vorhanden
ist, ausschliesslich vorhanden ist, das ist nicht nicht
vorhanden ist... Seitens dessen, was umfasst wird ...
liegt eine Umfassung vor, wenn die umfasste Beschaffenheit
ausschliesslich da, das ist an dem Beschaffenheitstra'ger,
an welchem auch die umfassende Beschaffenheit vorhanden
ist, vorhanden ist1.
6. The meaning of the term svabhava in Dharmakirti's work is
discussed in Steinkellner 1971 and 1974. As it is not easy
to find suitable English equivalents, I translate 'own
being1, svabhava, as 'essence1 and 'essential property'
thus following Steinkellner. The notion of svabhava plays
an important role in Dharmakirti's logic and epistemology.
The following remarks by Steinkellner may throw some light
on the import of svabhava, 'Aus diesen Stellen [ie from
PVSV] entnehme ich dass Dharmakirti den Svabhava als eine
Beschaffenheit (bhavah, dharmah) bestimmt, die nicht von
einer anderen Sache bedingt ist, d.h. mit der Sache selbst
(sva-) gegeben ist (Steinkellner 1971, p. 206)'. And fur-
ther, 'In the ontological context the svabhavah of some-
thing is the totality of causal possibilities, usually
determined only with regard to one major aspect, whereas
in a logical context the various aspects of the causal
possibilities have been separated by abstraction. If we
... translate "essential property, wesentliche
Beschaffenheit" in logical contexts, the translation
"essence, Wesen" would be quite appropriate in ontological
contexts...1, Steinkellner 1974. p. 124.
7. Cf the translation in Steinkellner 1971, p. 204, note 93,
113
'Welches [zunSchst] nichtwahrgenommene [Ding]/ das an sich
wahrnehmbar ist (tallaksanam), bei der Wahrnehmung
bestimmter [anderer Dinge] (yesam) [selbst] wahrgenommen
wird und nicht wahrgenommen wird, wenn unter diesen
[anderen Dingen] (tatra) ein einziges fehlt, das ist
dieses [Ursachenkomplexes] Wirkung'. Cf also Y. Kajiyama,
Trikapancakacinta, Miscellanea Indologica Kiotiensia, Nos.
4-5f 1963/ Kyoto University/ Kyoto, p. 2-3.
8. Cf Steinkellner 1971/ p. 185/ note 26 which contains the
following remark on this passage, 'Damit ist die ... Frage
... beantwortet/ wie man ... aus dem Rauch auf Feuer
schliessen kann. Feuer als Mitursache im Ursachenkomplexf
der allein den Rauch hervorbringt/ wird fflr sich allein
gar nicht erschlossen, sondern ist als logische Folge
(sadhyam) immer im Sinne eines Symbols fUr den
Ursachenkomplex zu verstehen, in dem die Mitursachen
Feuer, Holz, Feuchtigkeit/ Wind/ etc. vereinigt sind1.
9. Cf Steinkellner 1971, p. 202, 'Dharmakirti's Verdienst ist
es/ die Frage nach dem Grund f(lr den logischen Nexus
beantwortet zu haben. Und das Wesentliche seiner Antwort
ist/ dass er zeigt/ dass und wie der logische Nexus auf
einer realen Verbindung beruht und bei welchen Begriffen
diese Verbindung in der Wirklichkeit gegeben ist1.
10. These two verses are also translated in Steinkellner 1979,
p. 105 representing Pramanaviniscaya 11.60-61, 'Wenn der
TermitenhUgel das Wesen des Feuers (nMmlich Rauch
hervorzubringen) hatr dann ist er eben Feuer. Und hat er
nicht das Wesen des Feuers, wie kflnnte Rauch aus ihm
entstehen? (v.60) Das Feuer hat nSmlich das Wesen der Ur-
sache fflr Rauch und unterscheidet sich durch diese (Rauch
hervorbringende) F&higkeit (von anderen Ursachen).
Entstflnde der Rauch aus etwas, das nicht Ursache fflr Rauch
ist, dann wSre er ursachelos (v.61)1.
11. A simpler translation than mine is found in Steinkellner
1971, p. 205, 'Die reale IdentitSt (tadatmyam) dieser
114
Sache besteht nSmlich nur mit dem, was sich bloss an sie
anschliesst, nicht [aber] mit dem, was von einem anderen
abhMngt'.
12. Paksilasvamin seems to foreshadow Dharmakirti's
epistemological principle that the existence of a thing is
the perception of that thing.
••.pramanena sati qrhyamane tad iva yan na grhyate tan
nasti, yady abhavisyad idam iva vyajnasyata vijnanabhavan
nastlti tad evam satah prakasakam pramanam asad api
prakasayatiti ,
'When an existent [thing] is being perceived through a
means of valid cognition, than that which is in no way
(iva) perceived is not [present or not existent]. If this
[unperceived thing] would have been [present], it would
somehow (iva) have been cognized, [but] because there is
no cognition [of it], it is not present. Thus, a means of
valid cognition, while revealing what is present, reveals
also that which is not present (NBh, p. 4.2-4)'.
Paksilasvamin's last remark, that a means of valid cogni-
tion reveals present as well as absent things, formulates
a principle that seems to be implied in Dharmakirti's
theory on non-perception, anupalabdhi, as a valid probans.
13. In this verse Dharmaklrti alludes to a passage from
Dignaga's Nyayamukha. He quotes this passage in the prose
following on this verse (cf PVSV p. 18.17 and the critical
apparatus).
115
IV The Pramana-definitions of the Pramanasiddhi-chapter of
PV, v 1-7 (PV II.1-7)
In PV II Dharmaklrti exhaustively discusses the topics
that relate to the first half of PS I., a half-verse in
which Dignaga expresses his respect for the Buddha.1 PSV
I.1a-b runs:
pramanabhutaya jagaddhitaisine
pranamya sastre sugataya tayine /
'After paying obeisance to Him who is'a a means of valid
cognition, who strives for the welfare of the world, who
is the Teacher, the Well-gone, the Saviour1.
This verse praises the Buddha by simply enumerating five
of his outstanding virtues. The verse does not tell us
that the Buddha is a pramana on the grounds that he pos-
sesses the four other virtues. However, in the commentary
on this verse, Dignaga briefly explains why he thinks the
Buddha is a pramana, an authoritative speaker on religious
subjects2*
atra bhagavato hetuphalasampattya pramanabhutatvena
stotrabhidhanam sastradau gus pa bskyed par bya ba^i
don du"o
(PS 1.1 comm line 1-2), 'As regards this [verse], there is
at the beginning of this treatise a sentence praising the
Lord [Buddha] as one who is a means of valid cognition [or
religious authority] through [his] perfection in cause and
effect, in order to produce reverence [for the Buddha in
the hearers of this book]1. The perfection in cause (hetu)
and effect (phala) makes the Buddha the religious
authority that he is, makes him a pramana. In the rest of
the commentary, it is taught that cause refers to the
Buddha fs striving for the welfare of the world
(jagaddhitaisin) and his being the Teacher (sastr), while
effect refers to the Buddha's being the Well-gone (sugata)
and Saviour (tayin). Thus, according to Dignaga, the last
four gualities are meant to describe the nature of the
Buddha's religious authority .
tatra hetur asayaprayogasampat. asayo jagaddhitaisita.
116
prayogo jagacchasanac chastrtvam
(PS 1.1, comm line 2-3),
'With regard to this [foregoing statement, it is so that]
the cause is the [Buddha's] perfection in disposition of
mind and [his perfection in] practice. Disposition of mind
is [the Buddha's] being one who strives for the welfare of
the world. Practice is the fact that [the Buddha] is the
Teacher [by excellence] on the grounds that [he] teaches
the world'. In this way, the second and third virtue men-
tioned in PS I.1a-b are said to produce the cause of the
Buddha's authority.4 The effect is explained as follows:
phalam svapararthasampat. svarthasampat
sugatatvena trividham artham upadaya
(PS 1.1 comm line 3-4), 'The effect is the [Buddha's] per-
fection [in the virtues ] for himself as well as for the
sake of others. The perfection [in the virtues] for him-
self [is clear] from [his] being the Well-gone [which]
comprises three meanings'. These three meanings are the
following,
prasastatvam surupavat. apunaravrttyartham
sunastajvaravat. nihsesartham supurnaghatavat
(PS 1.1 comm line 4-5), '[namely] [the meaning of] (a)
being extolled like one who has got a beautiful form [ ie
body]^; (b) the meaning of not returning [anymore to the
world of birth and death the Buddha is called Well-gone]
like one whose fever has fully gone; (c) the meaning of
[his being] complete [ie perfect] like a jar that is com-
pletely full1. In the next sentence, Dignaga emphasizes
that the term sugata is an epithet of the Buddha alone.
don gsum po de yan phyi rol pa^i^dod chags dan bral ba
dan, slob pa dan, mi slob pa rnams las ran don phun sum
tshogs pa khyad par du bya ba^i phyir ro
(PS 1.1, comm line 5-6), 'And these three meanings [of the
term well-gone] are also [mentioned here] because [the
Buddha's] perfection [in the virtues] for himself should
be distinguished from [the achievements of] those non-
Buddhist [saints] who are freed from passion, [as well as
the achievements of] those [Buddhist saints] who are
[still] undergoing religious training, and those who no
117
[longer] need religious training [ie who are arhats]'.
This passage completes Dignaga's explanation of sugata.
The fifth epithet of the Buddha is his being the Saviour.
Dignaga says about this,
pararthasampaj jagattaranat tayitvam
(PS 1.1, comm line 6-7), 'The [Buddha's] perfection for
the sake of others is the fact that [he] is the Saviour
because he enables the world to cross [the waters of birth
and death]1. This continuous prose passage (PS 1.1, comm
line 1-7) gives a brief synopsis of the religious impor-
tance of the Buddha's personality. The Buddha has per-
fected and saved himself, but also saves others by teach-
ing them a road to ultimate salvation, a road which he has
travelled himself. This constitutes, according to Dignaga,
the Buddha's being a pramana.
Since Dignaga, in the Pramana-samuccaya, does not
elaborate on the relation of the concept of pramana with
the Buddha and his doctrines, it was left to Dharmakirti
to write and comment extensively in PV II on this verse
and its short commentary.
Even though we may say with Steinkellner that PS 1.1
'was conceived with a stroke of genius around 530 A.D.,
and can be considered the cornerstone that marks the his-
torical border between the dialectical and the truly epis-
temological period of our [ie Buddhist] tradition'
(Steinkellner 1982, p. 7), the verse still does not say
much more than that the Buddha is a pramana and possesses
four other qualities. The rest of the PS deals with the
two pramanas, perception and inference, as well as with
syllogistic reasoning, dialectics and semantics, but these
things are not related to the religious message of the
Buddha.6
Dignaga simply asserted that the Buddha is a means of
valid cognition and Dharmakirti considered it his task as
a commentator to explain this assertion. It presented to
him the following epistemological problem: Dignaga ac-
cepted only two means of valid cognition in PS I.2a-c; if
the Buddha is also a means of valid cognition, that is, if
his teachings are pramana, then his teachings must either
118
be regarded as inference or constitute a third kind of
pramana, namely reliable tradition (agama), in addition to
the accepted two pramanas. On the one hand, Dharmaklrti
does not simply posit tradition (agama) or reliable state-
ment (sabda, aptavada) as a separate pramana; on the other
hand, he does not regard the words of the Buddha as mere
inference , since they refer sometimes to things that are
not accessible to the perception and inference of the
hearers. But if the words of the Buddha give valuable non-
inferential information about the highest goals of
religious aspiration, it should be rationally established
to what extent these words are pramana.8 That it is neces-
sary to logically prove the trustworthiness (avisamvada)
of reliable statements on things that are beyond ordinary
perception (paroksa) is maintained by Dharmaklrti already
in PV I in discussing PS II.5a-b (see note 7).
tasya casyaivambhutasyaptavadasyavisamvadasamanyad
adrstavyabhicarasya pratyaksanumanagamye^py arthe
pratipattes tadasrayatvat tadanyapratipattivad avisam-
vado^numlyate
(PVSV, p. 109.7-9) ,
'And in so far as the statement by an expert [as taught by
Dignaga in PS II.5a-b] is of such a nature [as we have in-
vestigated in PV 1.214], is in general (-samanyad)
trustworthy, (analysis of the compound aptavadavisam-
vadasamanyad of PV 1.216a-b, then follows a loose
paraphrase:) it is inferred (anumlyate) that this [expert
statement] of which no error (vyabhicara) [in cases that
can be verified] has been observed is trustworthy also in
[the specific case of] an object that is [at present] not
accessible to perception and inference, on the grounds
that [also in the latter case] cognition [referring to the
invisible object and not only to the intention of the
speaker, cf PVSV, p. 109.10] has the [expert statement] as
its basis, [just] like cognition [referring] to [objects
that are] different from the [invisible objects, ie cogni-
tion referring to directly verifiable objects]1. Here
Dharmaklrti in fact acknowledges that information about
invisible things can only come from expert statements
119
(aptavada), but he demands that the truth of such informa-
tion should be inferred from the trustworthiness of the
information about visible things. In another passage from
the same work he says that, in the absence of other
pramanas, man needs reliable traditions (agama) in the
form of statements made by experts9 in order to know the
at-that-time invisible results of good and bad acts.
nayam puruso^nasrityagamapramanyam asitum samarthah.
atyaksaphalanam kesamcit pravrttinivrttyor
mahanusamsapapasravanat tadbhave virodhadarsanac ca
(PVSV, p. 108.2-5),1 Man cannot [safely] exist without resorting to the
[reliable] authority of tradition [consisting of reliable
statements], because [from reliable tradition, agama
alone] he hears the great advantage and the [great] disad-
vantage of engaging in or abstaining from certain [acts]
whose results are not [at present] perceivable, and [also
he cannot safely exist without reliable tradition] in so
far as he does not observe any contradiction [with the
reliable tradition] when the [results predicted by the
tradition] are [actually] there1. In PV II Dharmaklrti
says that the Buddha too made proper use of certain infor-
mation which he had got from reliable tradition, agama,
for example, information about the cause of suffering.
This is said in PV II.132cd-133ab, trsl Vetter, p. 40:
yuktyagamabhyam vimrsan duhkhahetum pariksate // 132 cd
tasyanityadirupam ca duhkhasyaiva visesanaih / 133 abf[Der Anwendung der Mittel geht voraus:] Forschend mit
Hilfe von Argumenten und Uberlieferung stellt er eine
Ursache des Leids fest und ausserdem, dass diese [Ursache]
eine nichtewige usw. Natur hat, [und zwar stellt er dies
fest] an Hand von Bestimmungen, die dem Leid selbst
[anhaften]'.
In PV II.1-7 Dharmakirti proposes to define and ex-
plain the first word of PS 1.1, pramanabhuta, in connexion
with the Buddha. In PV II.1-6 the term pramana is defined
and discussed, in PV II.7 the results of this discussion
are applied to the doctrines of the Buddha. What follows
now is a translation of PV II.1-7 with elaborate discus-
120
sion of its contents. In explaining the verses, I will
make use of the oldest and therefore most authoritative
commentary on PV II, Devendrabuddhifs Pramana-varttika-
panjika (PVP), preserved only in Tibetan.10 Devendrabuddhi
usually explains quite satisfactorily, but when certain
theories of Dharmaklrti are only hinted at in PVP, I ex-
plain them in greater detail by quoting relevant passages
from PV I which was written before PV II (cf Frauwallner
1954, p. 148, 152-153).11 The text of PV II I utilize is
PV I (Mi), so that PV II.1-7 corresponds to PV I (Mi) 1-7.
Below every part of the Sanskrit of PV II.1-7 I give
references to various other standard editions in which the
text of PV II occurs.
Devendrabuddhi introduces Dharmaklrti's general
definition of a means of valid cognition with the follow-
ing words:
tshad mar gyur pa zes bya ba ni tshad mar ^khruns pa^o //
tshad ma daft ^dra bas na tshad ma ste bcom ldan ^das so //
tshad ma^i no bo de ci zig yin na gan gis de dan ^dra bar
dam bca^ bar byed ce na
(PVP (P), p. 2a 1-3), '[In Dignagafs introductory verse it
is said that the Buddha] "has become a means of valid
cognition" which means fhe came into existence as a means
of valid cognition1. By being similar [in validity] to a
means of valid cognition, he is a means of valid
cognition, namely the Lord [Buddha]. [Question:] "What is
this nature of a means of valid cognition, that it is as-
serted that [the Lord Buddha] is similar to [a means of
valid cognition] ?".
The answer to this is given, according to
Devendrabuddhi, in Dharmaklrtifs definitions and their ex-
aminations offered in PV II.1-7.
1 a-b
pramanam avisaifivadi jnanam
(PVBh p. 3.21; P W , p. 3.14; P W (S), p. 3.5)
'A means of valid cognition [in general] is knowledge pos-
sessing trustworthiness.
The commentary of Devendrabuddhi says:
bslu ba yod pa ma yin pa can ni bslu ba med pa can te /
121
gan la bslu ba med pa zes bya ba^i don to //
(PVP (P), p. 2a,4-5), 'That in which untrustworthiness
does not occur is trustworthy; in which trustworthiness
occurs, [that is possessing trustworthiness], this is the
meaning [of Dharmaklrti's expression "possessing trust-
worthiness, avisamvadin"]'. According to PVP (P), p. 2a. 5-
2b. 1 it is possible to interpret Dharmaklrtifs expression
in two ways: (a) a means of valid cognition is knowledge
of a trustworthy thing, and thus knowledge possessing,
grasping, revealing the trustworthiness of the object; (b)
a means of valid cognition is knowledge which itself pos-
sesses trustworthiness, hence trustworthy knowledge. In
the first interpretation, trustworthiness is predicated of
the object of knowledge, in the second interpretation, of
knowledge itself. Although the second interpretation of
avisamvadin seems more natural and obvious, the first can-
not be wholly rejected as fanciful, for it occurs also
with Dharmaklrti himself in PV 1.215-217 and PVSV, p.
108.17-109.20 as we will soon see. It is conceiveable that
these passages have led Devendrabuddhi into thinking that
in PV II. 1a trust-worthiness is said to exist of the ob-
ject and the knowledge of it. He writes thus:
mi bslu ba de van don yons su bead nas "jug pa na don gyi
ran
no
de"
yul
q± nus
bo'i
i de
can
mi
lta
gyi
pa grub pas j
slu ba ni
bur qyur
chos yin
(D.
pa
no
i ltar "dod
p. 1b5; P.
rtoqs pa na
// de qan la
pa^i
'i>ses
yod
don
yul
pa'i
pa
de
qyi
mi
de
lta bu'i
chos
bslu
dan /
ba ni
ni bslu ba
med pa can gyi ses pa^o //
(PVP (P), p. 2a.5-2b.1),
'And further, this trustworthiness — when [on the part of
the perceiver] there is activity [directed towards a
thing] after the thing has been fully ascertained [by him
through means of valid cognition] — being the trustwor-
thiness of [this thing] whose form conforms to [the
thing's] desired purpose [ie desired and expected by the
perceiver] through the establishing [ie ascertainement] of
the thing's own power [to serve the desired purpose], is a
property of the object (yulf visaya) [of cognition]. And
when there is the cognition that has become [of] such [a
122
form, ie of the form] of the [thing], then the trustwor-
thiness of this knowledge is a property of [the valid
cognition] related to the object (yul canf visayin). This
[knowledge] in which the [trustworthiness of the object
and the trustworthiness of the cognition] is present, is
"knowledge possessing trustworthiness".
It seems that Devendrabuddhi interprets the double
trustworthiness in this manner; the power of a real thing
to serve an intended purpose for the perceiver constitutes
its trustworthiness, while the valid cognition of the
thing's power constitutes the trustworthiness of the cog-
nition of the thing. That is to say, we proceed from the
trustworthiness, the power of the thing, to the trustwor-
thiness of the cognition of the thing. Both forms of
trustworthiness eventually unite in the knowledge which is
the essence of a pramana.^
That trustworthiness (avisamvada) should be an essen-
tial quality of a pramana is not an entirely new doctrine
stemming from Dharmaklrti himself. For it is implied in
Dignaga's definition of a reliable statement by an expert.
aptavadavisamvadasamanyad anumanata /
(PS II.5ab)
'The statement by an expert is [to be regarded as] in-
ference in so far as there is similarity in trustworthi-
ness [between real inference and reliable statement]'.
Note that according to this definition aptavada and
anumana must share the essential quality of
trustworthiness, avisamvada (the same word as in PV II.1),
in order to be pramanas. Dharmaklrti evidently used this
term avisamvada to define the essential quality of pramana
in general in PV II.1ab. In PV 1.216-217 he interprets
avisamvada of Dignaga's definition in two ways, (a) as the
trustworthiness of the reliable statement and (b) as the
trustworthiness of the object of the statement; in this we
recognize Devendrabuddhi's two interpretations of the term
in PV II.1.
Before interpreting Dignaga, Dharmaklrti defines the
general characteristics of the expert statement;
sambaddhanugunopayam purusarthabhidhayakam /
123
parlksadhikrtam vakyam ato^nadhikrtam param //
(PV 1.214)
'A sentence [an expert statement]: (a) whose [words] are
coherent, (b) for [which] there are means that are
suitable [for acquiring the desired ends], and (c) which
expresses what is useful to man, is [alone] made the sub-
ject of an investigation [into the validity of such a sen-
tence as a pramana]. A [sentence] which is different from
[such] a [sentence having these three characteristics] is
not made the subject [of an investigation into its
validity]1. Only the statement of an expert which has
characteristics a-c could be regarded as a pramana,
provided the statement can be proved to be trustworthy.
kah punar asyavisaitivadah (PVSV, p. 108.16), 'But what is
the trustworthiness of this [useful sentence as defined in
PV 1.214]?'. Dharmaklrti answers:
pratyaksenanumanena dvividhenapy abadhanam /
drstadrstarthayor asyavisamvadas tadarthayoh //
(PV 1.215)1 The trustworthiness of this [useful sentence] about
visible and invisible things which are [ie can be] objects
of the [two pramanas, perception and inference], consists
in the fact that [the information contained in such a
sentence] is neither contradicted by perception nor by
twofold inference1.
'Twofold inference1, according to Karnakagomin, means:
anumanena ca dvividhena vastubalapravrttenaqamasritena ca
(PVSVT, p. 392.14-15), 'And "by twofold inference" [means:
by inference] that has operated through the power of a
[perceived real] thing and [inference] that is based on
tradition1. In PV 1.215 Dharmaklrti asserts that for an
expert statement to be trustworthy, two requirements
should be fulfilled — first, the information in the
statement must have been obtained by the speaker through
his own power of perception and inference? and second, the
information may not be contradicted by the subsequent per-
ception and inference of the hearer.
In the next verse Dharmaklrti uses this general con-
ception of trustworthiness to give his first explanation
124
of PS II.5ab.
aptavadavisamvadasamanyad anumanata /
buddher agatyabhihita parokse^py asya gocare //
(PV 1.216)
'In so far as a statement by an expert [ie by the Buddha]
is in general [ie in directly verifiable cases] trustwor-
thy [and thus reveals an object in an indirect manner as
does inference, therefore Dignaga] has designated [in PS
II.5ab] the cognition [derived from this statement] as in-
ference [ ie as of an inferential nature] also in [the
specific case, ie a case that is not directly verifiable,
of] an invisible object [to which] this [expert statement
refers], since there is no [other] possibility [but to
also regard the latter kind of statement as an inference,
ie in so far as no contrary results are perceived there is
no cause for deviating from Dignaga's rule that an expert
statement is like an inference]1.
In this first interpretation trustworthiness is
regarded as an essential quality which the statement as a
pramana and valid inference as a pramana have in qommon.
The second interpretation says that the expert statement
is inference in so far as the hearer knows that the main
object which the statement described is trustworthy.
heyopadeyatattvasya sopayasya prasiddhitah /
pradhanarthavisamvadad anumanam paratra va //
(PV 1.217)
'Or [introducing the second explanation of PS II.5ab, an
expert statement] about another [object, ie one that is
not visible to the hearer] is inference in so far as the
main thing [which the statement describes] is trustworthy
on the grounds that [this main thing, being the four noble
truths taught by the expert, ie the Buddha:] [namely] the
truth of what has to be avoided [ie suffering] and ob-
tained [ie the destruction of suffering] together with the
causes [ie the cause of suffering and the cause of the
destruction of suffering, the eightfold path], is estab-
lished [ie ascertained by the hearer to be trustworthy
through his own power of perception and inference]1.
In the prose following on this verse Dharmaklrti adds,
125
heyopadeyatadupayanam tadupadistanam avaiparltyam
avisamvadah
(PVSV p. 109.15-16), fThe fact that what is to be avoided
[ ie suffering] and obtained [ ie the destruction of
suffering] and the causes thereof [ie the cause of suffer-
ing and the cause of the end of suffering, the eightfold
path] which have [all] been taught by the [expert speaker,
the Buddha] are not contrary [to the perception and in-
ference of the hearer] constitutes the trustworthiness [of
these objects that have been indicated by the Buddha]1.
These excerpts from PV I and PVSV lend support to
Devendrabuddhifs interpretation of PV Il.lab that a means
of valid cognition in general is trustworthy knowledge as
well as knowledge of a trustworthy object, the difference
being that PV I refers to reliable statement and
inference, whereas PV Il.lab refers to every means of
valid cognition.
Devendrabuddhi also makes clear why the general
definition in PV Il.lab is valid for both perception and
inference, the only two pramanas recognized by Dignaga and
Dharmaklrti.
skyes bu ma slus pas ni mi bslu ba ni mnon par "dod pa"i
don dan phrad par byed pa"i mtshan nid can gyi mnon sum
dan / rjes su dpag par khyab par byed pa yin no //
(PVP (P), p. 2b.2), '[The term] "trustworthiness" [in the
general definition, meaning: trustworthy] "by not having
deceived a person [with regard to the intended effect of a
real thing]", comprises perception as well as inference [,
in so far as both pramanas are] characterized by the fact
that they lead to the intended [ie intended and expected
by the perceiver] effect [of a real thing]1.
In PV II.1bc Dharmaklrti defines the trustworthiness an-
nounced in PV Il.lab.
1bc
arthakriyasthitih /
avi saitivadanam
(PVBh, p. 4.1-2; P W , p. 3.18-19; P W (S), p. 3.7-8)
'Constancy [on the part of the thing and the cognition of
it] with respect to the production of a [useful] effect
126
[by the real particular thingf svalaksana] is the
trustworthiness [of the knowledge of the thing and of the
thing itself]f.
The reliability of the cognition of an object is
determined by the realness of the object. An object must
produce an effect (artha-kriya) in order to be real, and
if the object is constant in doing this, it is trust-
worthy. The cognition of this effect-producing object must
likewise be constant in order to be trustworthy cognition.
If it is true that Dharmakirti, in defining pramana,
predicates trustworthiness of both the object and the cog-
nition of the object in PV Il.lab, then the same trustwor-
thiness which is defined here in PV II.1bc must apply to
the object and its cognition. Thus sthitif "constancy" can
be consistently predicated of the object -- ie the real
object must be constant in its production of a (useful)
effect — as well as of the cognition of this object,
meaning that the cognition must be constant in its ac-
tivity of cognizing the production of an effect by the
real object. These two modes of constancy define the
trustworthiness, the reality of the object of cognition
and the truth, the pramanata, of the cognition of the ob-
ject respectively.
The definition in PV II.1bc is paraphrased by
Sakyamati in his commentary on PVP:
don ni sreg pa la sogs pa"o // de^i byed pa ni skyed pa"o
// de"i gnas ni rtogs pa ste khams kyi don sna tshogs pa
nid kyi phyir ro //
(PVT II (P), p. 88b.2), f "Effect" means [for example the
effects of a real fire like] burning [and useful effects
like cooking] etc. "Production" of this [effect] means
bringing [it] about. "Constancy" of the [production of an
effect by the real object] means the cognition [of the
effect-production], because the root [sjtha, to stand, from
which the noun sthiti, constancy is derived] has various
meanings' . The last remark seems strange, as the root
SPtha, "to stand" cannot mean "to cognize", but Sakyamati's
paraphrase is influenced by Devendrabuddhi's commentary
which says about PV II.1bc;
127
tshad mas yons su nes pa"i don gyis sgrub par bya ba^i don
byed par rtogs pa^o // ji ltar mnon sum gyis ses nas me
bsreg pa dan / "tshed pa la sogs pa^i nus pa la *jug pa
de"i dro ba^i reg pa la sogs pa^i yul can gyi mnon sum
dan ji ltar *ga* zig gi tshe me nid la rnam pa mtshuns pa
la sogs pa *ga* zig gis ^khrul pa du ba las me nes par
byas pa*i rjes su dpag pa^i *jug pa mi bslu ba yin no //
(PVP (P), p. 2b.3-5),
'The cognition of the production of an effect to be
brought about by a thing that is fully ascertained [by the
perceiver] through the means of valid cognition — like
[the cognition coming from] the direct perception refer-
ring to [effects] such as touching etc something hot when
[an object such as a real fire] is actually present —
[this perception is trustworthy cognition] with regard to
the power of fire to burn and [the conseguent useful power
to] cook [food] etc after [a real fire and its powers]
have been [correctly] known through direct perception; and
as the activity of inferring is trustworthy when at a cer-
tain time the [inference] has ascertained [the presence of
real] fire on the grounds that there is smoke [an in-
variably concomitant effect of fire with a smaller exten-
sion than fire] even though [the inference] is erroneous
with regard to the essence of [a real] fire because
[inference] operates with a [kind of] similarity of forms
etc [among real particulars, ie inference deals with con-
ceptualized and therefore ultimately unreal universals,
not with real particulars]1.
In this passage Devendrabuddhi maintains that it is
the cognition of a useful effect of a real thing, which
constitutes a pramana. The definition in PV II.1bc is dis-
cussed here in connexion with both pramanas. First, with
direct perception (pratyaksa), the ultimate source of all
knowledge — when perception of fire reveals some constant
effect to the perceiver, like heat, the power to burn and
thus the power to cook food etc, then these perceptions
are trustworthy in so far as they reveal to the perceiver
a real and trustworthy object. For a real fire will in-
variably produce such effects as burning and heat, which
128
can be used to cook dinner on it. Inference, on the other
hand, cannot show the real object, but it can and does
convey to the perceiver a general notion of the real
object, when the perceiver observes another real object
that is invariably concomitant with the first one and has
a smaller extension than the first one. The actual percep-
tion of smoke at a certain place conveys the general no-
tion of fire, with which smoke is invariably concomitant.
If the direct perception of the probans, smoke, is trust-
worthy, and if the invariable concomitance of the probans
with the probandum, fire, has been observed before in a
trustworthy way, then the inference of the presence of
fire will also be trustworthy, as the general notion of
fire which the inference conveys, can bring the perceiver
to the actual, real fire. The trustworthiness of both man-
ners of cognizing — through perception and inference —
of fire is warranted by the constancy of the production of
effects by the real fire. For fire is constant in produc-
ing actual heat as well as potential smoke. Even though
fire does not always produce smoke, when smoke is produced
it is only by fire, and thus smoke will invariably point
to the presence of real fire.
The notion of arthakriya, "production of an effect"
plays an important role in Dharmaklrtifs ontology.15 The
term did not originate with him, it occurs in older Bud-
dhist texts in the sense of "useful, beneficial action"
(Nagatomi 1967/68, p. 60) and it occurs with the same
meaning in NBh, p. 117.4; 445.3 (cf Nagatomi op cit p.
63). In Dharmaklrti's ontology however, arthakriya is the
essential power of a real thing, vastu, the particular,
svalaksana, visesa. In PV 1.166 together with PVSV, p.
84.4-11 Dharmaklrti defines the particular, real object
and its contrary, the conceptualized, unreal universal in
the following way.
sa paramarthiko bhavo ya evarthakriyaksamah / 166ab
idam eva hi vastvavastunor laksanam yad artha-
kriyayoqyata ^yogyata ceti vaksyamah.
sa ca
arthakriyayogyo^rthah nanveti yo^nveti
129
na tasmat karyasambhavah // 166cd
tasmat sarvam samanyam anarthakriyayoqyatvad avastu.
vastu tu visesa eva tata eva tannispatteh
(PVSV, p. 84.4-11),
'Only that which is able to produce an effect is a thing
which is real in the ultimate sense (PV I.166ab);
For this is the sole characteristic of a real thing
(vastu) and of that which is not a real thing (avastu),
namely the capacity [in the real thing] to produce an ef-
fect and the absence of this capacity [in what is not a
real thing], this we will explain.
And this object which is capable of producing an ef-
fect is not inherent [in other similar particulars as a
universal]. That which is inherent [in other particulars
as a seeming similarity, ie a universal] does not bring
about an effect (karya) (PV I.166cd).
Therefore, every universal is an unreal thing
(avastu), because [the universal is] not capable of
producing an effect. But only a particular (visesa) is a
real thing, because only the [particular, real thing]
brings forth the [effect]1.
This passage clearly demonstrates that arthakriya is
the essential power of a real thing (vastu)• Significantly
enough, the term is paraphrased as karyasambhava,
"bringing about an effect (karya instead of artha)" in PV
1.166d. Thus, valid cognition of a real thing and its
effect-production (arthakriya, karyasambhava) is not gen-
erated only by direct perception, but also by inference.
For the real thing that is not directly perceived is in-
ferred to be present on the grounds of perceiving an ef-
fect that is produced by, and therefore invariably con-
comitant with, the real thing that is its cause. This
logical reason is called karyahetu, "effect as probans".
Dharmakirti illustrates this kind of probans in PV 1.37:
dhumahetusvabhavo hi vahnis tacchaktibhedavan /
adhumahetor dhumasya bhave sa syad ahetukah //fFor fire [alone] possesses the essence of [being] the
cause of smoke [since fire alone is] in possession of a
particular power for the [production of smoke]. If the oc-
130
currence of smoke from [a thing that is] not the cause of
smoke [were possible], then it [ie smoke] would be without
a cause [, which is absurd}1. From this example of a
karyahetu/ it is clear that fire not only has the power to
produce effects such as heat which is useful for cooking
dinner, it can also, to quote here Devendrabuddhifs
example, produce an effect (arthakriya) such as smoke.
Smoke is a useful effect of fire when it indicates the
presence of a real fire.
In PV II.1cd Dharmaklrti applies his definition of
trustworthiness to meaningful, coherent verbal expressions
in order to make clear to what extent they can convey
valid knowledge to the hearer, ie how they function as
pramana.
1cd
sabde^py abhiprayanivedanat //
(PVBh, p. 4.21; P W , p. 4.11; PVV (S), p. 4.2)f[This trustworthiness] also occurs with [knowledge] con-
veyed by words, in so far as [these words forming a
coherent meaningful sentence] make known [to the hearer]
the intention [, ie the concepts in the mind of the
speaker]'.
Devendrabuddhi briefly explains this assertion,
sgra "byun van / bsam pa dag rii ston phyir ro (PV II.1cd)
// mi bslu ba yin te zes bya ba rjes su *jug go // skyes
bu^i bsam pa ston par byed pa de ltar na de ston pa^i
phyir / *di ni tshad ma nid yin no //
(PVP (P), p. 3b. 4), 'Also with [knowledge] conveyed
through words, in so far as [these words forming a
coherent meaningful sentence] make known the intentions
[ie the concepts in the mind of the speaker] (PV II.1cd),
there is trustworthiness, this [term] is [what should]
follow [in the verse]. The making known [through words in
a meaningful sentence] of the intention in a person ['s
mind] is a means of valid cognition in so far as making
known this [intention is done] in this way [ie by describ-
ing a trustworthy, real particular to the hearer]1.
A meaningful sentence can convey to the hearer
knowledge of a thing, namely knowledge of the intention
131
present in the mind of the speaker• When the knowledge
contained in the meaningful sentence leads the hearer to
directly perceiving a trustworthy and real object, then
the sentence itself is also trustworthy.
I understand sabda, "[knowledge] conveyed by words" as
referring to words used in a coherent, meaningful sentence
which teaches suitable means and expresses what is useful
to man, as such a sentence alone is regarded by Dharma-
klrti as a possible candidate for being a pramana. He made
this clear in PV 1.214 which I have quoted earlier.
Coherently formulated sentences which refer to practicable
methods to gain something and which, moreover, disclose a
desirable human goal (especially in the religious sense),
can really communicate practical knowledge about real ob-
jects to the hearer, even though such statements do not
reveal particular objects themselves as direct perception
does. For denotative words in sentences are mere mental
symbols of real objects. The use of informative sentences
is explained in PV 1.92-93.
sabdah samketitam prahur vyavaharaya sa smrtah /
tada svalaksanam nasti samketas tena tatra na // 92 //
api pravarteta puman vijnayarthakriyaksaman /
tatsadhanayety arthesu samyojyante" bhidhayakah // 93 //
'[Denotative] words express a [concept derived from real
objects, a concept] that is agreed upon [by linguistic or
philosophical convention]. This [agreement, ie linguistic
convention] is regarded [to be useful] for activity [on
the part of the hearer towards the real object, but] at
the time [of acting,] the [real object, being a momentary]
particular, is not present [to the perception of speaker
and hearer anymore], therefore the agreement [with regard
to the concept] is not [made with direct referece] to the
[momentary real particular] (92). [And yet] a person [ie the
hearer] who [— on the basis of words in a meaningful sen-
tence — ] has got [conceptual] knowledge of [real objects]
that are capable of producing [useful] effects, can act
for the accomplishment of these [for him useful effects];
thus [on the strength of linguistic convention] denotative
[words in a coherent meaningful sentence] are [fruitfully]
132
used to [point out] real objects (artha) [to the hearer]
(93) '.
Words do not express real things but the conceptual
pictures of real things as they are present in the mind.
This aspect of Dharmaklrtifs philosophical semantics will
be further elucidated in our comments on PV II.2. What
words denote in ordinary and philosophical or scientific
speech is something that is mutually agreed upon by the
users of these words. Words by themselves do not express a
thing; it is on the strength of a particular linguistic
convention (samketa) that denotative words put in a mean-
ingful sentence are made to express certain intentions
(abhipraya) or denote real objects by the speakers who use
these words. The intentions of the speaker are meaningful
constructions made out of concepts which are mentally
derived from perceived real objects, momentary particulars
(svalaksana). Therefore, these concepts are unreal; they
are not as real as the particulars from which they were
derived.
But these "unreal" concepts are very useful tools when
they can impart information about real objects that
produce useful effects.17 Such information is correct in
so far as the receiver, on the basis of it, strives for
and really acquires the desired effect-producing objects .
In PV II.2 Dharmaklrti makes clear what he means by
the "intention" of the speaker and defines the power of
words to convey valid knowledge to the hearer.
2
vaktrvyaparavisayo yo^rtho buddhau prakasate /
pramanyam tatra sabdasya narthatattvanibandhanam //
(PVBh, p. 7.23-24; PVV, p. 4.17-18; PVV (S) p. 4.3-5.1)
'[For the hearer] a [coherent meaningful] statement
(sabda) [coherently expressing suitable means to reach the
desired end and good human goals] is a means of valid cog-
nition [only] with regard to the thing which appears in
the mind [of the speaker] as the object of the [mental]
activity [ ie the intention] of the speaker, [but the
validity of such a statement] is not [in a direct manner]
based on the true [absolutely individual] nature of the
133
[real particular] object [ie the statement cannot reveal
the true nature of the particular]1.
As I have already shown, Dharmaklrti believes that
words in a meaningful statement do not reveal any real ob-
ject but only its mental image as it appears in the mind
of the speaker. A meaningful statement does not reveal the
true nature of the object, the real particular, for if it
really could, it would be equal to direct perception,
which it, by its very nature, is not. The trustworthiness
of perceptual cognition can be tested immediately, because
perception directly grasps the effect-production of the
particular. But the trustworthiness of a meaningful state-
ment is tested only afterwards, when the hearer really
perceives and acquires for himself the effect of the par-
ticular that was described in the statement.
Devendrabuddhi paraphrases this verse;
"chad pa po"i byed pa ste / van dag par "dod pa dan brjod
par "dod pa zes bya ba"i bar du"o // de"i yul gyi don gan
yin pa rab tu gsal zin (2b) so sor snan ba"o // gan la ze
na / rnam par rtog pa"i bio la"o (2b) // brjod par "dod
pa"i yul gyi don de la sgra "di ni tshad ma nid yin no
(2c) // "chad pa po"i don der snan ba can gyi rnam par
rtog pa sgra las rtogs par "gyur te / don de ses pa na "di
ni tshad ma nid yin no zes bya ba"i don to //
(PVP (P), p. 3b.5-7), '"The [mental] activity of the
speaker" means as much as [the speaker1s] real intention
and [his] desire to verbally communicate [the concepts he
has in his mind], the "thing which" is the object of the
[intention of the speaker] "appears" [meaning] becomes
manifest; in what [does the intention appear]?, [answer:]
in the conceptualizing "mind". "With regard to the" thing,
being the object of the intention [of the speaker], this
"[coherent meaningful statement] is a means of valid
cognition" (2c). Through the [meaningful] statement a con-
cept is cognized [by the hearer] which shows (pratibhasin)
the thing [ie the mental semblance of the real object] [in
the mind] of the speaker. [For the hearer] this
[meaningful statement] is a means of valid cognition with
respect to the [abstracted conceptual] knowledge the
134
[speaker has got] of the real thing [, the particular per-
ceived by the speaker]1.
In this passage, Devendrabuddhi makes it clear that
meaningful statements reveal only the concepts in the mind
of the person who is speaking, and such statements are in
a direct manner pramana only with regard to the concepts,
but not with regard to the actual objects which generated
the concepts.
Dharmaklrti has based his philosophical semantics on
the idea that words are symbols of concepts derived from
the real objects. Concept-making, in his opinion, is done
by the mind through its faculty of apohay "[mentally]
separating / excluding [a thing from other things]11. The
conceptualizing mind separates certain perceived objects
from other objects that are not quite like them, as well
as from those that are totally unlike them. Through this
separation / exclusion, the mind arrives at certain
(general) concepts. These concepts can be used to describe
real objects: either a single concept refers to a number
of objects or a certain number of concepts is made to
refer to a single object.
I will now quote a passage from PVSV which shows that
Dharmakirti regards words as images or symbols of concepts
(vikalpa):
nanu sarvato vyavrttasya rupasyabhidhanam na sambhavati /
na vai tad eva rupaffl buddhau samarpyate / anatlndriyatva-
prasangat / kevalam ayam tathabhutam pratyayayisyamlti
sabdena srotary asamsrstatatsvabhavam vikalpapratibimbam
arpayati
(PVSV, p. 37.23-27), 'Surely an [adequate] denotative ex-
pression for the form [of the momentary real particular
that is completely] separated [ie different] from all
[other momentary particulars] cannot exist nor indeed is
this very same form communicated to the mind [of the
hearer by a denotative expression], because [if it could
be communicated in its real essence] then it would wrongly
follow that [a word denoting a momentary particular] is
not above the senses [so that a denotative word could act
as if it were equal to the direct perception of the
135
denoted particular]. The [speaker] - while thinking "I
will make known [to the hearer a particular] that is of
such a [specific] nature" - through a denotative word [or
meaningful sentence] merely communicates to the hearer a
conceptual image with which the essence of the [momentary
particular itself] is not [directly] connected1.
Next, I will quote the famous passage with which
Dharmakirti starts his apoha-doctrine in PV I. this pas-
sage defines how, through apoha, correct concepts are
formed in the mind of the person who perceives a
particular.
sarve bhavah svabhavena svasvabhavavyavasthiteh /
svabhavaparabhavabhyam yasmad vyavrttibhaginah // 40 //
tasmad yato yato^rthanam vyavrttis tannibandhanah /
jatibhedah prakalpyante tadvisesavagahinah // 41 //
tasmad yo yena dharmena visesah sampratlyate /
na sa sakyas tato^nyena tena bhinna vyavasthitih // 42 //
(PV 1.40-42)
'As all things [, real, momentary particulars,] through
[their individual] essence, partake of [total] separate-
ness from their own being1^ [ie separateness from similar
things], and from another essence [ ie separateness from
unlike things] -- because [all particulars] abide
[exclusively] by their own essence — (40) therefore
specific classes which penetrate [ ie grasp] the specific
properties of the [real separate particulars] are [merely]
conceptually constructed [in the mind of the beholder];
[the concepts constituting these specific classes of
similar objects] are based on the [conceptual] exclusion
of every [unlike object] from the things [, ie from that
number of not so different particulars which in the mind
of the beholder produces the concepts of classes of
similar objects](41).
Therefore, through which [conceptualized general] property
[and words denoting this property] a special property [of
a particular] is known, [only] through this [general but
correctly circumscribing property], not through another
than this one, can the [specialness of the particular be
known and made known; for this [reason each concept has
136
its] separate existence [ ie its proper range of
applicability, so that qualitatively different concepts
referring to the same particular are not identical in
content] (42) .'
In verse 40 Dharmaklrti defines the uniqueness of the
momentary real particular: every particular is in reality
totally different from every other seemingly similar, as
well as dissimilar, particular. Every particular has its
own unique essence and therefore partakes of total
separateness from every other particular. In verse 41 he
described the mind's function of conceptual separation,
vyavrtti, apoha.
After particulars have been perceived, the mind concep-
tualizes specific classes of objects and certain general
properties. These are not based on the actual essence
(svabhava) of every real particular, but on some common
differences among real particulars. When a given number of
different particulars makes a somewhat similar impression
on the mind, it is possible to say that these particulars
share the same properties, and thus it is imagined that
they belong to a class of objects. But the similarity
among them is not based on their really sharing a real
universal by which they look similar; there is a seeming
similarity in them in so far as ' they are less different
from one another than from other more dissimilar
particulars. The mind envelopes the slight differences and
thus conceptually creates similarity. In verse 42 it is
said that the concepts which are thus created by mentally
excluding the too dissimilar particulars can be used to
denote (but never precisely) the special aspects of real
particulars. Concepts made by conceptual exclusion are
general and factually unreal, but those that correspond
most to the unique features of a particular have the
capacity to symbolize these unique features. Other con-
cepts that do not correspond very much to these features
cannot symbolize them. The idea that concepts and denota-
tive words circumscribe the particular only by excluding
from it all that is very dissimilar is unfolded in greater
detail in the following verses.
137
kvacid drste^pi yaj jnanam samanyartham vikalpakam /
asamaropitanyamse tanmatrapohagocaram // PV 1,48 //
'Also with regard to a [real particular] that has been
directly perceived somewhere and to which no parts [ie
properties] of other [particulars] have been [wrongly]
attributed, the conceptual knowledge [of it] has as its
object the [conceptualized] similarity [of things] and
operates in the sphere of merely excluding [from this
particular] the [wrong attributions] (48)'.
niscayaropamanasor badhyabadhakabhavatah /
samaropaviveke^sya pravrttir iti gamyate // PV 1.49 //
'Since an ascertaining [conceptual] cognition [of a
particular] prevents [wrongly attributing certain
properties to this particular] and since [whatever mode
of] cognition which [wrongly] attributes [these
properties] is what is to be prevented [by correct concep-
tual ascertainment, ie correct concepts], therefore the
[usefulness] is discerned [of the] operations of this
[ascertaining conceptual cognition] with regard to the
isolation of [wrong] attributions [from the mental image
of a real particular] (49)f.
In the next two verses Dharmaklrti equates the use of
denotative words with the correct conceptual ascertain-
ments (niscaya-manas) derived from exclusion,
yavanto^msasamaropas tannirase viniscayah /
tavanta eva sabdas ca tena te bhinnagocarah // 50 //
anyathaikena sabdena vyapta ekatra vastuni /
buddhya va nanyavisaya iti paryayata bhavet // 51 //
(PV 1.50-51 )1 There are just as many [conceptually constructed] ascer-
tainments [of a single particular] and [significative]
words [for these ascertainments] in order to expel [wrong
attributions to the particular] as there are [wrong] at-
tributions [to it] of parts [ie properties of other
particulars]. For this [reason] these [ascertainments and
words] have their various ranges [of applicability, ie
correct ascertainments and words have their own content
apart from the real particular they refer to, and since
this is so, they can be used to meaningfully refer to one
138
particular without being tautological] (50) .
Otherwise, if one single real [particular] object could be
[fully] comprehended [ie grasped in all its uniqueness] by
one [single] word or [one single instance of conceptual]
cognition, then there would be no other aspect [of the
particular left that is not expressed by this one concept
and word] and thus [all concepts and words referring to
this one particular] would be fully equal in meaning
[which in fact they are not]1 (51).
Significative words and the concepts they symbolize
do not grasp the real essence of a particular but only
remove wrong notions about a particular. It is possible to
entertain a great variety of wrong notions about a single
particular, and many words can be used to keep away these
wrong notions. Hence it is said 'there are just as many
ascertainments and words in order to expel [wrong notions]
as there are [wrong] attributions1. By removing as many
wrong imaginations about a particular as possible with the
help of ascertaining concepts and significative words, one
is not left with the real essence of the particular but
only with a useful delineation of, and a negatively-
defined boundary around, the particular. In verse 51
Dharmakirti gives a reductio ad absurdum: if it were true
that one conceptual ascertainment and one single word
could describe the absolute singularity of one given
particular, then all other words that denote this par-
ticular would carry exactly the same meaning as the first
word. But this is in reality not so, for we can and do use
different concepts and words such as "sweet", "heavy",
"cold" etc. to refer to and circumscribe one particular,
without these concepts and words being identical in
meaning, and these same concepts and words are used to
circumscribe other different particulars.
Regarding PV II.3ab I will first quote a passage from
PV I, which explains how the mind abstracts concepts of
things from the direct perceptions of real particulars.
According to Dharmakirti, the conceptualizing mind men-
tally envelopes or encloses (samvrti) the perceived, ab-
solutely singular appearance of the particular. In this
139
way, the absolute singularity is obscured by the mind and
the particular can be conceptually analysed into
substances, properties as well as put in classes of seem-
ingly similar objects. Thus, all notions of singularity,
classes and universals (samanya) are constructions of the
mind and do not have an existence independent of the mind.
According to Dharmaklrti, a universal is not a real thing
present in many particulars, but simply a convenient con-
ceptual image that can be used in inference or verbal com-
munication of knowledge.
pararupam svarupena yaya samvriyate dhiya /
ekarthapratibhasinya bhavan asritya bhedinah // 68 //
taya samvrtananarthah samvrtya bhedinah svayam /
abhedina ivabhanti bhava rupena kenacit // 69 //
tasya abhiprayavasat samanyam sat praklrtitam /
tad asat paramarthena yatha samkalpitam taya // 70 //
(PV 1.68-70)
'The [unique] form [of real particulars] which is
[absolutely] different [from the form of the concept] is
enclosed by the [conceptualizing] mind through [its
equalizing and generalizing] own form, [the conceptualiz-
ing mind] contains the [uniform] appearance of one object,
[although this seemingly uniform appearance] depends on
[various] separate [unique particular] things (68). These
things which are in themselves [absolutely] distinct, but
whose particularity has been [conceptually] enclosed by
the [conceptualizing mind which is] the enclosure, appear
[in the mind of the perceiver] in some [general] form or
other as if non-distinct [ie as if really similar] (69).
On account of the intention [ie attributing one uniform
appearance to various unique.particulars] [present] in the
[mind], the [seeming] similarity [of particulars] is
called real, but in the manner in which this [similarity]
is conceptually constructed by the [enclosure through the
mind] the [conceptualized similarity] is not real in the
highest sense [not as real as a particular] (70) f .
The mind's faculty of enclosing (samvrti) the unique
appearance of a particular and thus creating universals
(samanya) brings us back to PV II. In PV II.3ab Dharma-
140
klrti describes to what extent cognition produced by men-
tal enclosure is not a pramana.
3ab
qrhltaqrahanan nestam saitivrtam
(PVBhf p. 21.3; P W , p. 5.1; PVV (S), p. 5.2)
'[The conceptual knowledge present in the mind of the
speaker, which is] produced by [conceptually] enclosing [a
particular perceived by him, we] do not regard [as a means
of valid cognition for the perceiver nor for the hearer]
because it [only] apprehends [a blurred, conceptualized
image, a concept characterized in general by the inability
to produce an effect, this concept is derived from an
effect-producing particular] that has been [already fully]
apprehended [by the power of direct perception belonging
to the perceiver]1.
The internal mental cognition of concepts as a means
of valid cognition for the perceiver and the hearer of
statements about them is rejected here, because the con-
cepts that are apprehended by this conceptual knowledge
are ultimately unreal things that do not produce an
effect, unlike the particulars that do produce effects.
Verse 1ab teaches that only knowledge which possesses, viz
grasps the effect-production by real particulars may be
called pramana, which leaves no room for knowledge of con-
cepts as a pramana, precisely because concepts do not
produce a real noticable effect and cannot thereby warrant
their trustworthiness; or to put it bluntly: a real fire
gives heat, a concept of fire does not.
The adjective samvrta, 'produced by [mental,
conceptual] enclosure1, that is, produced by samvrti, ob-
viously refers to the samvrti-theory in PV 1.68-70,
quoted just before, consequently, Devendrabuddhi inter-
prets samvrta as 'knowledge of a particular as it is con-
ceptually enclosed and put in a category of similar ob-
jects (samanya); thus samvrta could also simply mean1 concept'• His commentary runs:
"di ltar bum pa dan / yod pa nid dan / grans dan *degs pa
la sogs pa^i yul can qyi kun rdzob kyi ses pa ni tshad ma
nid du mi ^dod do // ci"i phyir ze na / bzun ba *dzin pa^i
141
rgyu^i phyir ro // de la dan por mthon ba nid tshad ma nid
de "jug par byed pa yin no //
(PVP (P), p. 4a.1-2), •in this way, the knowledge which is
produced by [mental, conceptual] enclosure and which has
as its object [abstracted concepts] such as "pot" [or
potness], "existence", "number", "upward motion" etc is
not regarded as a means of valid cognition. Why not? Since
[this kind of knowledge] apprehends [in a conceptual man-
ner a particular] that has [already] been apprehended [by
direct perception], [therefore it is not a pramana]. With
regard to this [particular], only the first perception,
only this means of valid cognition, makes [the perceiver]
act1 .
Conceptual knowledge of a particular becomes
knowledge of abstract concepts like "existence", "number"
and "potness", but these concepts do not share the reality
of the particulars that generated them. The particulars
themselves and their effect-production can only be com-
pletely cognized through direct perception, which is thus
the sole source of every other mode of valid knowledge. °
Now one might ask to what extent conceptual knowledge is
useful in daily practice and therefore valid. Dharmaklrti
answers this in PV II.3bd.
3bd
dhipramanata /
pravrttes tatpradhanatvad dheyopadeyavastuni //
(PVBh, p. 21.23-24; P W , p. 5.1-2 PVV (S), p. 5.2-3)
'[Conceptual knowledge, ie knowledge of concepts produced
by the conceptualizing] mind is a means of valid cognition
[only] in so far as [succesful] activity [on the part of
the hearer] with regard to a real [effect-producing] ob-
ject that is to be avoided or obtained, has this
[knowledge formulated in a coherent, meaningful sentence
teaching suitable means to acquire the desired ends, as
well as expressing ultimate human purposes] as [its, ie of
succesful activity] basis1.
According to verse 3ab samvrtay the knowledge of a
concept is not a means of valid cognition, as it does not
grasp the effect-production of a real object. In this
142
passage, however, — as in verse 4ac — , it is not the
conceptual knowledge but the conceptualizing mindf dhl,
itself which is called a pramana. There seems to be a dif-
ference between the mind (dhl) and its products, the con-
cepts (samvrta). This apparent difference does not pose a
problem, because in Dignaga's and Dharmaklrti's epistemol-
ogy the pramana, here the dhl and its result, the pramana-
phala, here the samvrta, are not thought to be different,
savyaparapratitatvat pramanarti phalam eva sat /
(PS I.8cd)f[We say that] a means of valid cognition is [identical
with] the resultant [cognition and not different from it]
because [the resultant cognition is] cognized together
with the activity [of cognizing a thing through means of
valid cognition]1. This is what Dignaga says and what
Dharmaklrti accepts as principle.
Devendrabuddhi interprets dhl as conceptual knowledge
in his paraphrase of PV II.3bd.
blan bar bya ba dan dor bar bya ba^i dnos po^i yul can
dan / skyes bu^i don zes bya ba can gyi "jug pa de"i la /
de gtso bo nid kyi phyir te / ses pa gtso bo nid yin pa^i
rgyu^i phyir / bio ni tshad ma nid yin no //
(PVP (P), p. 4a.6-7), MI0f that activity" which refers to
(yul can) the real object that is' to be avoided or ob-
tained and is called "the object of human pursuit ", "in
so far as the [conceptual knowledge, formulated in
coherent, meaningful sentences which teach means that are
suitable for acquiring the desired goals of human
pursuit], is the basis [of this activity]", [this means:]
because [conceptual] knowledge is the basis [of this ac-
tivity on the part of the hearer], [therefore] the
[conceptualizing] mind [of the speaker] is a means of
valid cognition [for the hearer]1.
Assemblages of correct concepts do not reveal the
real individual nature of a particular, but they can, and
do, give some sort of useful information about a
particular. This useful information is conveyed through
coherent meaningful sentences to a hearer, who can after-
wards act in order to avoid or obtain the described
143
particular. This means that only statements on real par-
ticulars that produce useful effects are considered here.
The idea that a hearer of such statements can use the
statements as a kind of reliable pramana is found in PV
1,93 quoted before in my commentary on PV II.1cd.
The terms heyay 'what is to be avoidedf, and upadeya,fwhat is to be obtained1, are used in PV 1.217 and the
prose passage following on it, PVSV, p. 109.15 ff to
describe the scheme of the four noble truths taught by the
Buddha•
heyopadeyatattvasya sopayasya prasiddhitah /
pradhanarthavisamvadad anumanam paratra va // 217 //
'Or [introducing the second explanation of PS II.5ab, an
expert statement] about another [object, ie one that is
not visible to the hearer,] is inference in so far as the
main thing [which the statement describes] is trustworthy
on the grounds that [this main thing, being the four noble
truths taught by the expert, ie the Buddha:] [namely] the
truth of what is to be avoided [ie suffering] and obtained
[ie the destruction of suffering] together with the causes
[ie the cause of suffering and the cause of the destruc-
tion of suffering, the eightfold path] is established [,
ie ascertained, by the hearer to be trustworthy through
his own power of perception and inference]'.
In this verse heyopadeya and upaya refer to the four
noble truths whose trustworthiness can be directly
verified by direct perception and inference. If the four
noble truths, being the main object (pradhana) of the
statement, are proved to be trustworthy, then by inference
it is to be established that the invisible object of the
statement is also trustworthy. The prose more explicitly
relates heyopadeya to the four noble truths.
heyopadeyatadupayanam tadupadistanam avaiparltyam
avisamvadah / yatha caturnam aryasatyanafli vaksyamananltya
(PVSV, p. 109.15-16), 'The fact that what is to be avoided
[ie suffering] and obtained [ ie the destruction of
suffering] and the cause thereof [ie the cause of suffer-
ing and the cause of the end of suffering, the eightfold
path] which have [all] been indicated by_ the [expert
144
speaker, the Buddha] are not contrary [to the perception
and inference of the hearer] constitute the trustworthi-
ness [of these objects that have been indicated by the
Buddha]. As is the case with the subsequently explained
[ie in PV II) scheme of the four noble truths1. What is to
be avoided (heya) obviously refers to the first noble
truth that there is suffering, for suffering is what one
wishes to avoid. The cause (upaya) of heya, is the cause
of suffering, the second noble truth. What is to be ob-
tained (upadeya) is the destruction, the end of suffering
which is the third noble truth. The cause (upaya) of
upadeya, here the cause of what is to be obtained, the
cause of the end of suffering, is the fourth noble truth.
The term heyopadeya in PV II.3bd clearly alludes to these
passages in PV I and PVSV.
The notion of an object that is to be avoided or ob-
tained as the object of valid useful knowledge is older
than Dharmaklrti. We find it also in NBh, p. 6.1-2:
heyam tasya nirvartakam hanam atyantikam
tasyopayo^dhigantavya ity etani catvary arthapadani
samyag buddhva nihsreyasam adhiqacchati,
'What is to be avoided, the cause of what [is to be
avoided], the complete avoidance, and the means of this
[avoiding what is to be avoided] must be known; having
perfectly understood these four words [designating four]
facts [that must be realized] one obtains the supreme
good1. We should compare this assertion with another one
at the beginning of NBh:
pramanena khalv ayam jnata^rtham upalabhya tarn
artham abhlpsati jihasati va
(NBh, p. 2.2), 'When a cognizer has cognized an object
with a means of valid cognition, he either wishes to ob-
tain (abhlpsati) that object or to avoid (jihasati) it1.
Paksilasvamin uses the verbal forms that correspond seman-
tically to the verbal adjectives heya and upadeya. About
the object one wishes to obtain or avoid, Paksilasvamin
says:
arthas tu sukhaiti sukhahetus ca duhkhaiti duhkhahetus ca
(NBh, p. 3.1), 'As to the object, it is happiness and a
145
cause of happiness, as well as suffering and a cause of
suffering1. To my mind, it is possible to relate NBh, p.
6.1-2/with this passage. We get the following relation:
sukhay happiness, corresponds to hana atyantika, complete
avoidance of suffering; sukhahetu, the cause of happiness,
corresponds with upaya, the means of avoiding suffering;
duhkha, suffering, is heya, and is to be avoided;
duhkhahetu, the cause of suffering is tasya nirvartaka,
the cause of what is to be avoided. Without much
difficulty, we may recognize the four noble truths of Bud-
dhism here. According to Paksilasvamin, the avoidance of
suffering and the obtainment of happiness is achieved on
the basis of correct knowledge acquired through the
pramanas. In this way, he wishes to prove the practical
value of the pramanas. For almost the same purpose
Dharmakirti seems to use the same terms heya and upadeya
in PV II.3bd, in order to show the practical value not of
every pramana, but especially of reliable statement, for
the practical value of the conceptual knowledge contained
in the statement, the being a pramana of conceptual
knowledge, consists in the fact that it gives the hearer
of the statement useful information about things that
should be avoided or obtained. If we ask ourselves why
Dharmakirti uses these terms, heya and upadeya, in PV
1.217 and PV II.3bd, the answer could be that these two
terms can be used to describe in a short basic formula the
kind of objects regarding which the pramanas have practi-
cal value, and the formula also defines briefly the
central doctrine of the Buddha, namely the four noble
truths. As we have seen, Dharmakirti interprets heya and
upadeya in this way in PVSV, p. 109.15-16, thus making
quite explicit the covert Buddhism of NBh, p. 6.1-2; p.
2.2; p. 3.1.
The mind is not only a means of valid cognition on
the grounds that it can produce useful concepts that give
images of useful or unuseful real objects, but is is also
a means of valid cognition when it correctly grasps the
unique features (cognized through direct perception) of
the real objects. This is explained in PV II.4ac.
146
4ac
visayakarabhedac ca dhiyo^dhigamabhedatah /
bhavad evasya tadbhave
(PVBh, p. 22.20; p. 23.9? P W f p. 5.10; 5.15; PVV(S), p.
5.4-5)
'And [the conceptualizing mind of a reliable speaker is
also a means of valid cognition for the hearer] in so far
as the [conceptualizing] mind (dhl) has a specific cogni-
tion (adhigama) [of the particular] owing to the specific
form of the [real] object [ie the particular], because the
[specific mental cognition of the particular] is only
present [in the conceptualizing mind of the speaker], when
the [specific form of the particular] is present [before
the perceiver who is the speaker after he has conceptual-
ized the specific form of the directly perceived par-
ticular and communicates the concepts of it to the
hearer]f.
Dharmaklrti says here that the specific form in which
a directly perceived object appears in the mind cor-
responds to the specific form of the object as it has been
grasped by direct perception (pratyaksa). And yet, the
conceptual image of the object is not equal to the real
proper form of it. Even if the image presents a very good
likeness, it still remains an incorrect copy of the
original particular. Dharmaklrti makes this clear in PV
1.87:
samrjyante na bhidyante svato^rthah paramarthikah /
rupam ekam anekaffi ca tesu buddher upaplavah //
'Objects that are real in the highest sense [ie the real
momentary particulars] are by [their] own [nature] neither
united [in order to form a class of real universals], nor
are they [, each individually, further] divided [into sub-
stances and properties]. [Thus,] a single or manifold form
[which seems to be present] in these [particulars] is a
disturbance [ie an error] of the [conceptualizing] mind
[ie the single or manifold appearances of the particulars
are simply created by the mind]1. As has been explained in
PV 1.68-70, the specific mental image of a particular is
created in the mind by conceptually enclosing (samvrti)
147
the perceived appearance of the real particular. The con-
ceptualizing mind of the speaker makes the semblance into
concepts and words which are spoken to the hearer. In so
far as the verbal communication about the particular
enables the hearer to find the particular for himself, is
it a pramana; not because the communication reveals the
particular itself. The conceptualizing mind itself (namely
the mind of the speaker) is a pramana in so far as it has
more or less correctly conceptualized the particular. For
the correctness of the conceptualization by the mind is
the basis of the correctness of the statement that is af-
terwards made regarding the conceptualized particular. Yet
the solid basis of all these mental conceptual operations
remains direct perception (pratyaksa), for perception
alone can give birth to correct concepts abstracted from
actually perceived particulars. All correct concepts
derive from the actual perception of particulars and from
nothing else. By demanding that perception alone must be
the source of accurate concepts (as is done here in PV
II.4c), Dharmaklrti can warrant the truth, the being a
pramana of the mind that conceptualizes. The mind can only
be a source of valid cognition for others if it apprehends
and conceptualizes really perceived and existant
particulars. Imaginations and fancies which could also be
derived from the direct perception of things are here not
considered to be pramana.
4d-5a
svarupasya svato gatih //
pramartyam vyavaharena
(PVBh, p. 25.2-3; P W , p. 5.20-p 6.15? PVV (S), p. 6.2-3)
The cognition (gati) [by the conceptualizing mind] of the
proper form [of cognition as such takes place] by the self
[-awareness of the mind's cognition], [but the cognition
by the mind] is [known to be] valid cognition [only]
through activity [based on this cognition and directed
towards the cognized particular]1.
Devendrabuddhi paraphrases PV II.4d as follows;
raft rig pa las tshad ma zes bya ba"i ses pa de^i ses pa
yod pa nid grub pa yin gyis / tshad ma nid ni ma yin no //
148
(PVP (P), p. 5b.5-6),'Even though by the self-awareness
[of the mind's cognition] the cognitive nature (ses pa yod
pa nid) of this cognition which is called "means of valid
cognition" is established, [still the cognition by the
conceptualizing mind] is not a means of valid cognition1.
Devendrabuddhi interprets svatah as ran rig pa las,
svasamvedanat, 'mental self-awareness'. This theory that
mental cognitions cognize themselves and do not need
another inner witness to be mentally perceived, is
Dignaga's theory of mental self-awareness, svasamvedana,
which is formulated in PS 1,9 and its commentary (Hattori
1968, p. 183.18-19). This passage has been discussed in
the chapter on Dignaga (p. 59).
Along with Devendrabuddhi, we could ask ourselves:
"o na ji ltar tshad ma nid nes par bya ze na
(PVP (P), p. 6a.1), 'And how is it to be ascertained that
[the cognition by the mind] is a means of valid
cognition?'. The answer to this question is given in PV
II.5a, pramanyam vyaharena, which Devendrabuddhi paraph-
rases thus,
dus phyis ^byun ba can gyi don byed pa^i yul can gyi ses
pas so
(PVP (P), p. 6a.1-2), '[The cognition by the conceptualiz-
ing mind is a means of valid cognition only] through the
cognition which refers to (yul can) the [actual] produc-
tion [by the particular] of the [expected] effect which
takes place at a later time.1 This means that every cogni-
tion is a pramana after the effect-production of a par-
ticular is actually perceived. Before that happens, there
is no criterion by which one can ascertain the validity of
a cognition. This would imply that coherent statements
that form a philosophical or scientific reliable treatise
(sastra) really serve no purpose, since their validity as
pramana is proved only afterwards, not immediately.
Devendrabuddhi introduces Dharmaklrti's solution of this
epistemological problem with these words:
gal te tha snad kyis tshad ma nid rtogs pa yin pa de^i
tshe bstan bcos don med pa can yin pa"i phyir brtsam par
bya ba ma yin par *gyur ro // don med pa can ma yin te
149
qan qi phyir
(PVP (P), p. 6a. 2-3), 'If the validity [of the cognition
by the mind] is known [only] through activity [based on
this cognition and directed towards the cognized
particular], then, since a [meaningful scientific]
treatise [made up of sentences mediating such cognition]
is useless [as its validity can only be established after
the indicated particular has really been experienced],
[such a treatise] should not be composed [at all].
[However, Dharmaklrti says that a scientific treatise] is
not useless, because...1.
5b
sastram mohanivartanam /
(PVBh, p. 29.19; PVV, p. 7.18; PVV (S), p. 7.2)
[The correct conceptual knowledge concerning particulars,
formulated by a reliable speaker in a coherent scientific
or philosophical] treatise [only serves to] dispel the
confusion [or ignorance of the hearer regarding these
particulars]f.
A scientific treatise is a set of coherent meaningful
sentences — ideally in the form of valid syllogisms, al-
though this is not always necessary — which convey more
or less correct and useful conceptual images of a real ob-
ject to the hearer. A treatise does not reveal the object
itself, it only describes the useful effects the object is
expected to produce. Thus, the validity of the treatise
depends on the actual effect-production of the denoted
particular. The real use of the reliable treatise lies in
its power to remove the confusion and ignorance in the
mind of the hearer. For the hearer is capable of forming
some more or less correct notions, ie correct mental
pictures, of a particular on the strength of the informa-
tion from the treatise. Devendrabuddhi gives a reason for
this description of the purpose of a sastra:
"jug par byed pa^i tshad ma^i mtshan nid mi ses par "jug
pa na bslu ba srid pa"i phyir ro //
(PVP (P), p. 6a.3), '... because errors are possible if
one acts without knowing the definition of the means of
valid cognition as inducing [the hearer] to act [in order
150
to get the expected particular and its intended effect-
production] ' *
The prime use of a treatise or a reliable tradition
is, of course, its information about real objects that the
hearer does not at present perceive for himself. The
treatise or tradition can incite the hearer to act in or-
der to acquire the at present invisible thing. This is
pointed out by Dharmaklrti in PVSV, p. 108.2-5:
nayam puruso 'nasrityagamapramanyam asitum samarthah,
atyaksaphalanam kesamcit pravrttinivrttyor
mahanusamsapapasravanat tadbhave virodhadarsanac ca,
'Man cannot [safely] exist without resorting to the
[reliable] authority of tradition [consisting of reliable
statements], because [from reliable tradition alone] he
hears the great advantage and the [great] disadvantage of
engaging in or abstaining from certain [acts] whose
results are not [at present] perceivable, and [also he
cannot safely exist without reliable tradition] in so far
as he does not observe any contradiction [with the reli-
able tradition] when the [results predicted by the
tradition] are [actually] there1. In the last clause,
Dharmaklrti describes the validity of reliable tradition
(agama), which must be the same as the validity of a
scientific treatise (sastra), as both must necessarily be
composed of meaningful, coherent sentences. Tradition and
a scientific treatise are both valid in so far as the pre-
dicted results, the predicted particulars and their
arthakriya, are afterwards experienced by the hearer.
Dharmaklrti now proceeds with a statement which
grammatically looks like a continuation of the assertion
sastram mohanivartanam, but which is interpreted by
Devendrabuddhi and all his successor commentators as an
alternative, second general definition of pramana.
5c
ajnatarthaprakaso va
(PVBh, p. 30.2; PVV, p. 8.1; P W (S), p. 8.2)
'Or [perceptional and conceptual knowledge in order to be
a genuine pramana not only possesses trustworthiness, but
also] reveals a [particular] object that was [previously]
151
not known [ie not cognized by any pramana]1.
Devendrabuddhi explains this passage as follows:
qzan mtshan nid gnis pa yin no // rtogs pa pos mi ses pa^i
don gyi dnos po^i de kho na nid kyi gsal bar byed pa ste /
mnon par gsal bar byed pa^i ses pa yan tshad ma yin no //
don byed par mi bslu ba nid kyi phyir de dan de rtog par
bya'o //
(PVP (P), p. 6b.1-2), 'This is another, second definition
[of pramana]; that which reveals an object, [ie] the
reality of a thing (dnos po^i de kho na nid) that was
[previously] not cognized by [the person] who [now] cog-
nizes [it]: cognition which clearly reveals [such an
unknown object] is also a means of valid cognition. This
[knowledge which reveals an unkown thing] must be inves-
tigated on the grounds of [its] trustworthiness with
regard to the effect-production [by the particular that is
revealed by this knowledge] together with that [cognition
which is trustworthy]'. Dharmaklrti in his definition, and
Devendrabuddhi in his explanation of it, demand that a
pramana, apart from being trustworthy, should reveal sound
knowledge to the perceiver, of particulars that were
unknown to him before he cognized them. In short: a
pramana should produce new, valid knowledge. This epis-
temological requirement as formulated in PV II.5c comple-
ments the definition of a pramana given in PV II.1ab. We
might say that a genuine means of valid cognition should
reveal an effect-producing particular that was not known
before. With the second definition in PV II.5c Dharmaklrti
seems to justify the possibility of continuously acquiring
more and more new, reliable knowledge of reality. If we
apply this definition to the individual pramanas, we find
that, in fact, it only refers to direct perception, for
only through direct perception is it possible to directly
cognize new objects, previously unknown particulars.
Through inference it is not possible to directly cognize
new objects, but only previously unknown relations between
objects. For example, we infer the presence of a par-
ticular A at a certain spot because that spot possesses
particular B which is known to be invariably concomitant
152
with particular A. If PV II.5c is applied to reliable
statement, tradition (agama) and a scientific treatise
(sastra), it seems to define them as well in a way. For it
is tenable to say that a correct statement describes an
effect-producing particular that the hearer of the state-
ment has never perceived for himself. Thus, a statement
can also reveal a previously unknown particular. This in-
terpretation of PV II. 5c as referring to inference and
reliable statement as well does not contradict the defini-
tion in PV II.1cd that the trustworthiness of a pramana
occurs with knowledge conveyed by words, in so far as they
make known to the hearer a correct conceptual image with
which he can search for the corresponding effect-producing
particular. However, since sound, verifiable knowledge of
particulars can only be had through direct perception
which has a direct relation with particulars, it is jus-
tifiable to interpret PV II.5c as defining direct
perception. The validity of the other pramanas is based on
the validity of direct perception, for only through direct
perception of a thing is the validity of inference and
reliable statement made certain.
According to Devendrabuddhi, the use of the term
"artha", the real object, in PV II. 5c implies that
Dharmaklrti wants to exclude false appearances, illusory
objects from being appropriate objects of valid cognition.
don smos pas ni zla ba gnis la sogs par snan
ba tshad ma nid ma yin par bsad de
(PVP (P), p. 6b.2-3), fSince he speaks of "objects"
[Dharmaklrti] has explained that [cognition of unreal,
false] appearances such as two moons etc, is not a means
of valid cognition1. Cognition of illusory objects is not
a means of valid cognition, because such unreal things —
even though they may have been previously unknown to the
perceiver -- do not, and cannot, produce any useful ef-
fects (arthakriya) for him.22 Devendrabuddhi further
argues that the word artha means the presence or absence
of a particular:
don smos pas kyan dnos po^i yod pa dan med pa
nid kyi de kho na nid yin par bzed do
153
(PVP (P), p. 6b.4), 'By saying "object" [Dharmaklrti] also
assumes that [this object] is the reality (de kho na nid)
of the presence and absence of a thing1. A means of valid
cognition can reveal the presence of an effect-producing
particular. In both cases, the knowledge is useful.
Through direct perception it is possible to directly know
the presence of a thing, for instance of a real fire;
through inference we know only indirectly the presence of
fire, namely through the concomitant smoke (if there is
smoke). When a particular is not directly cognized through
pratyaksa, we can infer the absence of this particular.
This inference constitutes -- as we will know -- Dharma-
klrti' s third kind of valid probans (hetu) , namely the
anupalabdhi-hetu, the non-perception of a perceivable ob-
ject as the logical reason to infer the absence of this
object. Devendrabuddhi has this probans in mind when he
says that the pramana as defined in PV II.5c reveals the
hitherto unknown presence of a particular, as well as the
hitherto unknown absence of a particular. By saying this,
Devendrabuddhi makes this definition refer to direct per-
ception and the valid inference as taught by Dharmaklrti
in PV I.23
After giving his second definition of a real pramana,
a definition primarily referring to direct perception,
Dharmaklrti goes on to show how this second definition
would affect the possible validity of the conceptual
knowledge of universals.
5d-6a
svarupadhiqateh param / /
praptam samanyaviinanam
(PVBh, p. 30.28-29; PVV, p. 8.17-18; PW (S), p. 84-5)
'Following upon cognition [through perception] of the
proper form [of the pa r t i cu l a r ] , the [conceptual]
knowledge of the [seeming] similarity (samanya) [of the
perceived object with other objects] would be (prapta)24
[a genuine pramana, which i t is not according to our
definition in PV II.5c in so far as conceptual knowledge
does not reveal any hitherto unknown aspects of a
particular, that would not have been perceived through
154
perception]'.
By rejecting conceptual knowledge of similarity as a
pramana, Dharmakirti is asserting here that the source of
all concepts (when they are correct, ie corresponding
somewhat to the appearance of the real particular) is the
direct perception of particulars. For the conceptual
knowledge of similarity (samanyavij nana) is abstracted
from the appearance of particulars. Conceptual knowledge
of an object is acquired only after (param) the cognition
(adhigati) of the particular, or rather the proper form
(svarupa) of the particular. Devendrabuddhi paraphrases
verse 5d-6a as follows:
ran gi mtshan nid khofl du chud pa las dus phyis spyi^i
rnam par ses pa gan yin pa de mnon sum gyis ma bzun ba nid
kyi spyi yul nid du bdag gir byed pa^i phyir / de yan
tshad ma nid thob na mi *dod do
(PVP (P), p. 6b. 5-6), fIf -- after the cognition [through
perception] of the particular -- the [conceptual] cogni-
tion of the [seeming] similarity [of this particular with
other particulars] is acquired [ie is regarded] as a means
of valid cognition, in so far as it is assumed that this
[conceptual cognition] has as its object a [special]
similarity which cannot be apprehended through perception,
[then we say that this] is not intended [ie no conceptual
knowledge of similarity is pramana]1. The reason that con-
ceptual knowledge is rejected here as a means of valid
cognition must be the fact that conceptual knowledge does
not reveal anything new about the perceived particular (cf
in this connexion also what was said in PV II.3a).
All mental abstractions in the form of universals,
and the similarity in a certain number of objects, are
made by the mind through its power of apoha, vyavrtti y ex-
cluding wrong concepts from a perceived particular. What
Dharmakirti briefly refers to in PV II.5d-6a, namely ac-
quiring knowledge of similarity, he explains more fully in
PV 1.87-88:
samrjyante na bhidyante svato "rthah paramarthikah /
rupam ekam anekam ca tesu buddher upaplavah // 87 //
bhedas tato^yam bauddhe^rthe samanyam bheda ity api /
155
tasyaiva canyavyavrttya dharmabhedah prakalpyate // 88 //
'Objects that are real in the highest sense [ ie the real
momentary particulars] are by [their] own [nature] neither
united [in order to form a class of real universals] nor
are they [each individually further] divided [into sub-
stances and properties]. [Thus,] a single or manifold form
[which seems to be present] in these particulars] is a
disturbance [ie an error] of the [conceptualizing] mind
[ie the single or manifold appearances of the particulars
are simply created by the mind] (87) '.
Therefore, also this [seemingly real] difference
[between] "similarity" [among various particulars] and
"particularity" [of one particular] is [in reality only
made] with reference to an object which is conceptualized
[by the mind]. Specific properties of this one [single,
real particular] are conceptually constructed [in the mind
merely] by way of excluding other [wrong attributions from
this real particular] (88)'.
In the ontology of Dharmaklrti, reality is built up
exclusively of evanescent momentary particulars (cf also
the chapter on Dignaga). From this assumption it follows
that every notion a perceiver might have of a lasting
similarity, a universal, inherent in an indefinite number
of particulars, is an erroneous mental construction of the
mind. But also the lasting particularity of a single ob-
ject is a mental construction since the single particular
itself changes every moment, even though this may not be
noticed by the perceiver. The only pramana that could
really convey a true notion of the particular is direct
perception, for as long as the noticeable effect of the
particular (arthakriya) is conveyed, direct perception is
the only verifiable pramana. Conceptual knowledge, on the
other hand, lacks this verifiability as it does not ap-
prehend the actual effect-production.
In PV II.6bc Dharmakirti gives the reason for not
regarding conceptual knowledge as a means of valid
cognition.
6bc
avijnate svalaksane /
156
yaj jnanam ity abhiprayat
(PVBh, p. 31.3-4; PVV, p. 8.21-22; PVV (S), p. 9.1-2)
'Because [our] intention [with the definition in PV II.5c
is] that [only sound, perceptual] knowledge of a
[hitherto] unknown particular is [a genuine means of valid
cognition]'.
Evidently, Dharmaklrti is merely paraphrasing the
definition offered in PV II.5c. Devendrabuddhi rephrases
PV II.6bc as follows:
^khrul pa med pa kho na yin no // ran gi mtshan nid ma
rtogs pa^i yul can gyi ses pa gan yin pa de ni tshad ma
yin pa de ltar na / khyad par dan bcas pa^i mtshan nid
bsad pas spyi ses pa ni tshad ma ma yin no //
(PVP (P), p. 6b.8-7a.2), f[The definition given by
Dharmakirti in PV II.5c] is certainly (kho na) non-
erroneous [in so far as it only refers to the apprehension
of the effect-producing particular]. [Only] knowledge
which refers to a [hitherto] unknown particular is a means
of valid cognition; by thus explaining the definition [in
PV II.5c] with a [limiting] specification, [Dharmakirti
makes clear that conceptual] knowledge of similarity is
not a means of valid cognition1.
In conclusion, we might ask along with Devendrabuddhi
why only knowledge of hitherto unknown particulars is a
means of valid cognition.
ci"i phyir ran gi mtshan nid rtogs pa^i ses pa gan yin pa
de tshad mar ^dod kyi gzan ma yin ze na
(PVP (P), p. 7a.4), 'Why is [only] knowledge consisting in
the cognition of a particular regarded as a means of valid
cognition, but not [any] other [form of knowledge]?1 The
answer to this is given in PV II.6d.
6d
svalaksanavicaratah //
(PVBh, p. 31.4; PVV, p. 9.5; PVV (S), p. 9.3)
'on the grounds that [a person who seeks to obtain useful
things] examines [with the means of valid cognition]
[real] particulars [which alone are capable of producing
the intended useful effects]1.
157
Reality is built up of particulars, and a person who
needs what is useful can only strive to obtain particulars
which produce an effect that meets his demands. The
criterion of truth of the pramana is, therefore, its being
helpful in making the perceiver obtain the desired effect
of a particular. If knowledge is not instrumental in gain-
ing the useful effects of a thing, then that knowledge is
not a trustworthy pramana. Devendrabuddhi says about PV
II. 6d
ran gi mtshan nid kho na don gyi bya ba byed pa de ltar na
de^i don du skyes bu ran gi mtshan nid don byed pa^i rten
^dod pa yul de nid la tshad ma tshol bar byed kyi / don
byed pa ma yin pa^i yul gzan la ni ma yin no //
(PVP (P), p. 7a.4-5), 'Only a particular brings about the
production of an effect; thus, for the purpose [of gaining
what is useful and avoiding what is damaging,] a person
who regards the particulars as the basis of the effect-
production, searches for a means of valid cognition con-
cerning only that [effect-producing] object, but not con-
cerning other objects, [ie other than particulars, namely
the conceptualized universals] that do not produce any ef-
fects [because the universals are not real in the highest
sense]1. With this observation of Devendrabuddhi's, we can
round off the discussion on the two pramana-definitions of
Dharmaklrti. We might summarize the epistemological posi-
tion of the latter by saying that a true means of valid
cognition should be verifiable knowledge of real objects
that produce certain perceivable effects which are useful
for the perceiver. Such knowledge is verifiable only in so
far as the effect-production by the particulars is ac-
tually experienced (PV II.1). In order to be a source of
new information about reality, a means of valid cognition
should reveal particulars that were unknown before (PV
II.5c).
In PV II.1-6 Dharmaklrti has discussed pramana in
general, thus defining the first word of PS I.1ab. He then
applies the results of this discussion to the religious
authority of the Buddha.
158
7a
tadvat pramanam bhagavan
(PVBh, p. 32.4; PVV, p. 9.11; PVV (S), p. 9.5)
The Lord [Buddha] is a means of valid cognition such as
that [which we have discussed in PV II.1-6]1.
Dharmaklrti implies here that the teachings of the
Buddha constitute trustworthy knowledge and reveal to the
hearer facts that were not known to him before. This im-
plication is clearly set forth by Devendrabuddhi in his
commentary.
*di tshad ma dan yan ci zig mtshuns na / gan gis tshad ma
dan *dra bas tshad ma nid yin / tshad ma^i mtshan nid rnam
pa gnis nid dan mtshuns pa yin no // ji ltar mnon par *dod
pa bzin du sgrub par bya ba^i don la mi slu ba^i don la mi
slu ba"i phyir dan / mi ses pa*i don gsal bar byed pa"i
phyir tshad ma nid yin no // de ltar na bcom ldan *das
kyan nes par legs pa la sogs pa*i mtshan nid can gyi skyes
bu*i don la mi slu bar mdzad pa dan / rtogs pa po"i skyes
bu^i don bsgrub par bya ba mi ses ston par mdzad pa de^i
phyir tshad ma nid yin no //
(PVP (P), p. 7b.2-5),
'In what way is the [Buddha] equal to a means of valid
cognition, that through similarity to a means of valid
cognition He is a means of valid cognition? He is equal to
the two ways [in which] a definition of a means of valid
cognition [has been given in PV Il.lab and 5c
respectively]. In accordance with what has been admitted
[in PV II.1-6 about a genuine pramana, the Buddha] is a
means of valid cognition in so far as [He] is trustworthy
with regard to the [useful] thing that is to be realised,
and in so far as [the Buddha] reveals [ie points out to
others] a [real] thing that has been unknown [to them].
Thus, also the Lord [Buddha] does not deceive (avisam-
vadayati, mi slu bar mdzad pa) with regard to [ultimate]
goals of human [aspiration, goals] defined as the highest
good (nes par legs pa, nihsreyasa) etc and [the Lord
Buddha] explains to a person who understands [Him] objects
of human [interest] that have to be realised [by the
hearer] and that are not [as yet] known [to the hearer];
159
therefore [the Buddha] is a means of valid cognition [as
defined in PV II.1-6]1.
The trustworthiness of the Buddha is proved by the
extent to which a person really acquires for himself the
objects taught by the Buddha. The Buddha's teaching is
trustworthy when the hearer finds the objects that con-
stantly produce desired useful effects (arthakriyasthiti)•
This is the trustworthiness with regard to visible things,
things that are accessible to the perception and inference
of the hearer. The Buddha's trustworthiness about in-
visible things must be inferred from his trustworthiness
about visible things. This has been set forth by Dharma-
klrti in PVSV, p. 109.15-19:
heyopadeyatadupayanam tadupadistanam avaiparltyam
avisamvadah / yatha caturnam aryasatyanam vaksyamananltya/
tasyasya purusarthopayogino^bhiyoqarthasyavisamvadad
visayantare^pi tathatvopagamo na vipralambhayanuparodhat,
nisprayojanavitathabhidhanavaiphalyac ca vaktuh,
'The fact that what is to be avoided [ie suffering] and
obtained [ie the destruction of suffering] and the causes
thereof [ie the cause of suffering and the cause of the
end of suffering, the eightfold path] which have [all]
been indicated by the [expert speaker, the Buddha] are not
contrary [to the perception and inference of the hearer]
constitutes the trustworthiness [of these objects that
have been indicated by the Buddha]. As is the case with
the subsequently explained [ie in PV II] scheme of the
four noble truths [ie suffering, the cause of suffering,
the destruction of suffering and the path leading to the
destruction of suffering]. Because this [teaching] of the
[Buddha about trustworthy facts that are accessible to the
perception and inference of the hearer] is trustworthy [in
so far as this teaching] is conducive to [ultimate] human
goals and [in so far as this teaching] is capable of being
put into practice [by the hearer], [therefore] one [must]
admit the truth [of this teaching] also [when it is
dealing] with another object [that is not accessible to
the perception and inference of the hearer], [the
hearer] is not deceived [with this second kind of
160
teaching] since [the first kind about visible facts] is
not disproved [by his own perception and inference] and
[the second kind of teaching is true also] because a
[reliable] speaker [of the greatness of the Buddha] does
not gain [anything] by [uttering] untrue statements that
[do] not [set forth human] aims1.
As for the invisible object that is indicated in the
teachings of the Buddha, it seems that this sort of object
will, in the course of time, have to become visible to the
hearer. For if it can never be experienced at all, we
could not with certainty call it existent. Moreover, the
Buddha himself at some time should have experienced the
object that is invisible at present, if he is to sig-
nificantly talk about it. According to the definition in
PV II. 1 meaningful, coherent statements, in order to be
pramana, should refer to particulars that are constant in
producing useful effects. But also statements about in-
visible things should ultimately refer to effect-producing
particulars, not to things that remain absolutely
imperceptible. It seems, therefore, that Dharmaklrti with
his definition in PV II.1 introduces a more effective and
explicit means of verifying the truth of statements about
invisible things than the theory in PVSV, p. 109.15-19.
According to the latter passage, the invisible object
could theoretically remain invisible forever and still the
hearer is obliged to believe in the existence of this in-
visible object, only because the reliable expert, ie the
Buddha, who described the invisible object, does not
profit by uttering false statements. According to PV II. 1,
however, a reliable statement about invisible objects
should explicitly refer to real particulars that are only
at present not perceptible to the hearer but will be so in
the future, when the hearer actually perceives them.
As we have seen, Devendrabuddhi applies in his com-
mentary the definitions of PV II.1 and 5c to the teachings
of the Buddha, and says that the Buddha's doctrine reveals
facts that the hearer did not know about before. This idea
is not entirely new, because already Paksilasvamin says
that the reliable speaker reveals facts that his hearers
161
did not know themselves. The hearer could only know these
facts from the correct instruction of the reliable
speaker, the expert (apta);
tesam khalu vai pranabhrtam svayam anavabuddhyamananaiTi
nanyad upadesad avabodhakaranam asti
(NBh, p. 223.7-8), 'Now, when these [ordinary] living
beings do not know for themselves [through their own power
of cognition those things that should be avoided or
obtained], then there is no other cause of [correct]
knowledge [left for them] than the [valid] instruction
[given by the expert speaker]1. It is very well possible
that this statement influenced Devendrabuddhi when he
wrote to the effect that the teachings of the Buddha
reveal hitherto unknown facts to the hearer. Of course in
Devendrabuddhi's view, the ideal and most elevated expert
is the Buddha. Even Dharmakirti may have been influenced
by this NBh-passage, as he maintains in PV II.7a that the
Buddha is a means of valid cognition equal to the sort of
pramana he has defined in PV II. 1-6, thus implying that
the words of the Buddha evince trustworthiness and revela-
tion of new facts.
After explaining the term pramana in connexion with
the doctrine of the Buddha, Dharmakirti explains the term
bhuta, "has become" of PS 1.1a.
7 be
abhutavinivrttaye /
bhutoktih
(PVBh, p. 32.8-9? PVV, p. 9.15; P W (S), p. 9.6-7)
'In order to [make his readers] dismiss [the notion that
something] that has not become [, ie something eternal
like the Vedas or a Supreme God, could be a real means of
valid cognition, Dignaga uses] the expression "has become"
[in PS I.1a, thus saying that the Buddha is only a pramana
because he has become so as the result of a process of
becoming],1
The Buddha has become an authoritative speaker on
religious subjects because he has trained himself over
many lives to become a fully enlightened one, a Buddha.
The Buddha's authority is the result of a process of
162
growth, of becoming. By saying "has become" Dharmaklrti
makes Dignaga reject the idea that an eternal supreme
being or everlasting Vedic utterances could be regarded as
a means of valid cognition. In PV II.8-9 Dharmakirti
rejects the possibility of the existence of an eternal
pramana.
Yet the process of becoming alone does not warrant a
person's being a means of valid cognition; the authorita-
tive speaker must have practiced certain moral excellences
which are the conditions and cause of his authority. On
the grounds of the speaker's possessing the required moral
excellences, the hearer can validly infer the authorita-
tiveness of the statements made by the speaker, as will be
made clear later. Devendrabuddhi writes about 7bc:
bcom ldan ^das kyi ye ses gan yin pa de dan de ni skyes pa
yin gyi / no bo nid kyis grub pa ni ma yin no // gan gi
tshe skye ba yin pa de^i tshe yul dan dus dan ran bzin gyi
nes par mi run ba^i phyir de ni glo bur bar rigs pa ma yin
no zes don gyis (PVP (D), p. 7a.2 gyi) ran gi rgyu ston
par byed do //
(PVP (P), p. 7b.6-7),
'That which is the higher knowledge (ye ses) of the Lord
[Buddha], even though all that (de dan de, tattat) has
come into existence, [yet it] is not perfected (grub pa,
siddha) [as pramana] [merely] through its own nature. When
[this higher knowledge of the Buddha] comes into being,
then — since it is not correct that [this higher
knowledge] is limited (nes par, niyata) to [some
particular] place, time and nature (ran bzin, svabhava)
[but developed over many lives] — this [knowledge] is not
suitable for being [regarded as merely] incidental (glo
bur bar, akasmikam P W , p. 10 note 1) [ie as something
that emerged by accident, not through conscious effort
over an enormous period of time]. Thus, he [Dharmakirti]
shows [in PV II.7cd], according to the [implied] sense,
the proper cause (raft gi rgyu, svakaranam PVV, p. 10 note
1 ) [why the higher knowledge of the Buddha has become a
pramana]'.
163
PV II.7cd
sadhanapeksa tato yukta pramanata //
(PVBh, p. 32.9; PVV, p. 10,2; PVV (S), p. 9.8)
'The validity of the Buddha as a pramana] depends on
[specific] means, therefore [the Buddha] is an appropriate
[ie correct and reliable] means of valid cognition1.
Devendrabuddhi says about this passage:
tshad ma de ni bcom ldan "das kyis // sgrub byed la bltos
pa (sadhanapeksa) / gan gi phyir bcom ldan "das tshad ma
nid du bzugs pa dan / de"i sgrub par byed pa dag "chad par
"gyur ro // de"i sgrub par byed pa nan tan du byed pa las
kyan tshad ma nid yod par "gyur ba de"i phyir (tato) /
bcom ldan "das ni tshad ma nid (pramanata) yin par rigs so
(yukta) // sbyor ba ni gan gis sgrub par byed pa gan phyin
ci ma log par nan tan du byas pa des ni / de thob par
"gyur te / dper na nad pas nad med bsgrub par byed pa
phyin ci ma log pa nan tan du byas pa lta bu"o //
(Skt fragment in PVV p. 11 note 1:)
yo yatsadhanam aviparitam anutisthati tasya tatpraptir
bhavati / yathaturasyarogyasadhanam aviparltam
anutisthatah. bcom ldan "das kyan tshad ma nid kyi sgrub
par byed pa phyin ci ma log pa nan tan mdzad pa can yin no
zes bya ba ni ran bzin gyi gtan tshigs so /
(PVP (P), p. 7b.7-8a.2) (cf Skt fragment PVV, p. 11 note
1:)
anusthitapramanyaviparltasadhanas ca bhagavan iti
svabhavahetuh,
'The validity [of the Buddha as pramana] depends on
[specific] means' [developed] by the Lord [Buddha]. On
which [grounds] the Lord [Buddha] is at present a means of
valid cognition and the means for [becoming] that, will be
[elaborately] explained [in the sequel of PV II]. Because
[the Buddha] too undertakes [certain intentional] actions
with regard to [acquiring] the means for the [higher
knowledge], [His] being a pramana could come into
existence; therefore the Lord [Buddha] is an appropriate
means of valid cognition. The formulation [of the proof in
PV II.7cd] [runs] as follows: He who undertakes [specific]
action (anutisthati, nan tan du byas pa) with regard to a
164
means [for obtaining] something [, in so far as this means
is] correct [ie leading to the intended results], obtains
that [for which he practiced the means], just as for a
sick person who undertakes [specific] actions with regard
to the means for [regaining his] health [, in so far as
this means is] correct [ie really leading to health],
[health will come back]. The Lord [Buddha] too is one who
has undertaken [specific] action with regard to a correct
[ie leading to the intended results] means for [becoming]
authoritative [ ie for becoming a means of valid
cognition]. This is [an inference in which] the essential
property ["correct means" is used] as probans .
The implied svabhavahetu-inference Devendrabuddhi is
formulating in the last part of this passage can be con-
structed as follows.
Two essential properties are involved; the first, with the
wider extension, is "practicing a means (sadhana) which
invariably leads to the intended results (aviparita)"; the
second, with the smaller extension, is "practicing this
means with the specific purpose to become an authority
(pramana)". The logical connexion between these two es-
sential properties, ie their invariable concomitance, is
based on the fact that both are forms of specific inten-
tional actions (anusthana, nan tan). Thus, it is possible
to infer the epistemological authoritativeness (pramanata")
of the Buddha from His having practiced correct means
(sadhana), since the latter by being the wider essence of
the former, invariably includes the former. Devendrabuddhi
had introduced the notion of intentional action (anustana)
in order to warrant the invariable concomitance of
pramanata with sadhana and the possibility of inferring
pramanata from sadhana, for sadhana has a wider extension
than pramanata. The only other possibility would have been
to regard pramanata as an invariably concomitant effect of
sadhana (which it in reality seems to be), but then one
could not validly infer the presence of pramanata as ef-
fect from the presence of sadhana, the hypothetical cause
°f pramanata. For, according to Dharmaklrti's logic, the
inference of an effect from its cause is not conclusive.
165
In PV II.34a (Mi) Dharmaklrti tells us the means
(sadhana) of the validity of the Buddha: sadhanam karu-
nabhyasat, 'The means [by which the Lord Buddha has become
a means of valid cognition] is [his great29] compassion
[on all living beings; this compassion is the solid ground
of his authority] because [he has without end] practiced
[his compassion]1. The Buddha's practice of compassion in
many lives became the foundation of his authority, of his
now being a means of valid cognition, because his inten-
tional effort at being compassionate is not limited to one
life, but it can bef and is, made over an enormous period
of time. For this reason, it becomes the solid basis of
the epistemological truth and trustworthiness of a Buddha.
In another passage Dharmaklrti says about the Buddhafs
compassion:
tayah svadrstamarqoktir vaiphalyad vakti nanrtam /
dayalutvat parartham ca sarvarambhabhiyoqatah // 145 //
tatah pramanam
(PV II.145-146a trsl Vetter, p.52)fDas Beschtitzen (taya) besteht im Verkilnden eines Pfads,
den er selbst [als zum Ziel der Leidlosigkeit filhrend] er-
fahren hat. [Er ist also sachkundig.] Er verkilndet [aber
auch] nicht [mit Absicht] Unwahres; denn [filr sich selbst
braucht er] nichts mehr zu erreichen und er tut alles, was
er unternimmt, da er ganz von Mitleid erf (lilt ist, urn an-
derer willen. Daher ist er Erkenntnismittel'.
In this verse Dharmaklrti sums up the religious and
epistemological authority of the Buddha and his doctrine.
The Buddha has realized for himself (svadrsta) the supreme
goal; he teaches (ukti) the way (marga) to this goal in a
reliable manner because he would not profit by making
untrue statements (anrta), for there is not anything he
has to gain (vaiphalya). This disinterestedness is caused
by his great compassion (dayalutva), and it is out of com-
passion that the Buddha acts, lives and teaches (sarvaram-
bhabhiyoga) in order to promote the supreme welfare of the
world (parartham). This statement by Dharmaklrti on the
validity of the Buddha strongly reminds us of Paksila-
svamin's three criteria of the reliability of an expert
166
speaker:
kim punar aptanam pramanyam saksatkrtadharmata
bhutadaya yathabhutarthacikhyapayiseti
(NBh, p. 223.4-5), 'Now, what is the [epistemological]
validity of the expert [speakers]? It is the fact of
having directly perceived the dharma [the true state of
things], compassion on living beings, and the desire to
communicate the thing as it really is1. Now Dharmaklrti's
svadrstamarga corresponds to Paksilasvamin's saksatkrta-
dharmata (cf about this term Oberhammer 1974, p. 50),
Dharmakirtifs vakti nanrtam, "the [Buddha] does not utter
untrue [statements]11, is similar to Paksilasvamin' s
yathabhutarthacikhyapayisa, while both regard compassion
as a criterion of validity (dayalutva in PV 11.145 and
daya in the NBh). It is very likely that Dharmaklrti more
or less adopted Paksilasvamin's three criteria, at the
same time giving compassion priority over the other two in
order to emphasize the specialness of the Buddha's
authority. For according to Dharmaklrti, the first cause
and driving force behind the Buddha's religious and epis-
temological authority is his compassion, as an attentive
reading of PV II trsl Vetter will show (consult the index
of that book on karuna, krpa, daya). Paksilasvamin1s list
suggests a temporal order of the criteria. The authorita-
tive speaker first has perceived a thing or fact himself,
then he is compassionate, and out of compassion he seeks
to correctly describe the thing or fact to others.
aptah khalu saksatkrtadharmana idam hatavyam idam asya
hanihetur idam asyadhigantavyam idam asyadhigamahetur iti
bhutany anukampante. tesam khalu vai pranabhrtam svayam
anavabuddhyamananaiti nanyad upadesad avabodhakaranam asti
... hanta vayam ebhyo yathadarsanaiti yathabhutam upadisamas
ta ime srutva pratipadyamana heyam hasyanty adhigantavyam
evadhigamisyanti iti evam aptopadesah
(NBh, p. 223.5-224.3), 'The experts having directly per-
ceived the dharma [the true state of things], show compas-
sion on living beings in the following way: this is
[something] to be abandoned, this is the cause of the
abondonment for him [ie for the living being the expert
167
wishes to instruct], this is [something] to be attained by
him, this is the cause of the attainment for him. For,
when these [ordinary] living beings do not know for them-
selves [through their own power of cognition those things
that should be avoided or obtained], then there is no
other means of [correct] knowledge [left for them] than
the [valid] instruction [given by the expert speaker]...
[the compassionate experts think:] Let us instruct these
[helpless living beings about the things that should be
avoided and obtained] in the way [we have] perceived
[these things for ourselves] and according to how [these
things] really are. When these [helpless living beings]
have heard [our accurate instructions], then, while
[correctly] understanding [them] [ ie in so far as they
correctly understand the instructions], they will avoid
what is to be avoided and attain what is to be attained;
in this way the instruction by the experts [is a
pramana]•'
The real difference between Paksilasvamin and
Dharmaklrti seems to be the following: Paksilasvamin does
not say how or why the apta has directly perceived the
true state of things (saksatkrtadharmata); the apta simply
has acquired his direct experience of higher things. Out
of compassion, he relates his experience and knowledge to
others and therefore his reliable statements can be
regarded as pramana (cf also Oberhammer 1974, p. 50 ff on
this). Dharmaklrti, on the other hand, does tell us how
and why the apta par excellence, the Buddha, has acquired
his direct experience of important religious facts; the
Buddha was moved by compassion and therefore exerted him-
self to attain the complete freedom from suffering in or-
der to teach others the way to that same goal. Before he
reached the goal himself, the Buddha had heard from reli-
able tradition (agama) about the cause of suffering, so
that he knew what to strive for, namely the ultimate end
of suffering for the sake of teaching to others the way to
it. Dharmaklrti says this in these two verses.
daya~v5n duhkhahanSrtham upgyesv abhiyujyate //
paroksopeyataddhetos tadakhyanam hi duskaram //
168
(PV II.131cd-132ab trsl Vetter, p. 39)
'[Nach langer tlbung nun] voll Mitleid wendet er [ie the
Buddha] die Mittel an, urn [sein eigenes] Leid zu
tiberwinden [obwohl er als einer, der voll Mitleid ist, die
eigene Erltisung gar nicht mehr anstrebt, Dennoch ist es
ntitig, die Mittel anzuwenden, urn das eigene Leid zu
flberwinden;] denn wer das Ziel [der Erlflsung vom Leid] und
die Mittel dazu nicht selbst erfahren [und angewendet] hat
(paroksa), dem dilrfte es kaum gelingen, sie [d.h. Ziel und
Mittel] [anderen gut] zu verkflnden1,
yuktyaqamabhyam vimrsan duhkhahetum parlksate //
tasyanityadirupam ca duhkhasyaiva visesanaih /
(PV II.132cd-133ab trsl Vetter, p. 40)
'[Der Anwendung der Mittel geht voraus:] Forschend mit
Hilfe von Argumenten und Oberlieferung stellt er eine Ur-
sache des Leids fest und ausserdemf dass diese [Ursache]
eine nichtewige usw. Natur hatf [und zwar stellt er dies
fest] an Hand von Bestimmungen, die dem Leid selbst
[anhaften]'.
169
NOTES
1. Cf Frauwallner 1954, p. 143; Nagatomi 1959, p. 264, 266;
Vetter 1964, p. 32; Hattori 1968, p. 73-74.
1a. Vetter interprets pramanabhuta as fder Erkenntnismittel
istf in Vetter 1984, p. 14 and note 6 on the same page.
2. It seems that Dignaga is deliberately playing with two
connotations of the term pramana; the first is of course
"source of valid knowledge" in a purely epistemological
sense, the second more general connotation is "authority",
"right measure" (cf Burrow 1980 on the etymology of
pramana). If pramana, in both connotations, is predicated
of the Buddha it would mean that he has become an expert
or authority on religious questions of ultimate concern to
mankind and that his statements on religious matters could
safely be regarded to b£ as valid as the knowledge derived
from perception and inference. Cf also Hattorifs remark on
pramanabhuta, fThe term "pramanabhuta" is used in a double
sense. First it means "authoritative" or "standard"...
Second, it has the more technical meaning, "to have to
come into existence" (bhuta) as a "means of valid
cognition" {pramana)f, Hattori 1968, p. 74, note 1.2.
3. Cf the scheme in Hattori 1968, p. 74, note 1.2.
4. Cf Steinkellner 1982, p. 8, 'The fact that the Buddha has
become a means of valid cognition is the result of the
development of certain qualities to perfection. The
qualities are further differentiated as perfections in
cause (hetu) and perfections in effect (phala), where the
two terms "cause" and "effect" bear the meaning they have
as categories of describing the career of a Bodhisattva'.
5. Cf Hattori 1968, p. 75, note 1.5: "Mahavastu, I, 92,13 and
Avadanasataka, 1,188,1 ff, relate the story of Surupa, a
legendary king, who, in exchange for religious
instruction, gave up his son, his wife, and himself to be
170
eaten by an ogre. His religious ardor is praiseworthy.
However, here "surupa" is to be taken as a common noun ac-
cording to Durvekamisra, who states, in explaining
"prasastata", that those who make a living by their beauty
of form are called surupa; see DhP, p. 3.15: surupa
rupajIvah . "
6. Steinkellner seems to object to what he calls11 secularistic" interpretations of the Buddhist epis-
temological tradition (Steinkellner 1982, p. 6-7). He dis-
agrees with certain modern scholars who praise the Bud-
dhist epistemological tradition (as founded by Dignaga
with the Pramanasamuccaya) 'highly for the very reason
that it is assumed to present a development of rational
secularization within Buddhist monastic culture ... that
it presents dawning of "modern, Western" progress within
the "dark ages" of traditional religious India1
(Steinkellner 1982, p. 6). According to Steinkellner, the
'secularistic1 understanding by modern authors of Buddhist
epistemology is mistaken as it represents this epis-
temological tradition as 'an essential deviation from the
spirit of Buddhism1 (op cit, p. 6). The fact remains that
the Pramanasamuccaya can very well be regarded as a
'secularistic1 work on epistemology (only two means of
valid cognition are accepted, ie perception and
inference), syllogistic reasoning, dialectics and
semantics, since -- apart from PS I.1ab — the book does
not link its theories to Buddhist spirituality. Of course
Dignaga did write about Buddhist spirituality, but not in
the Pramanasamuccaya. Thus, it is not hard to understand
why some modern scholars, reading the Pramanasamuccaya (as
well as a good many works of Dharmaklrti), have come to
the abovementioned judgement that Buddhist epistemology is
'secularistic1 and even 'modern'. One could only say that
these modern scholars show a too great positive bias
towards what they consider to be 'western progress1. In
the Pramanasamuccaya the only link with the Buddhist
religion is PS I.1ab and the commentary on it by Dignaga
himself; therefore Steinkellner has to say that 'its [ie
171
of PS I.1ab] importance is not only underlined by the
literary masses written as a commentary to it, but also by
the attitude toward it in later Buddhist history1 (op cit,
p. 7). To me it seems that this 'important1 verse (or
rather half-verse) is made important by the earliest and
most important commentary on it, Pramana-siddhi chapter,
being Pramanavarttika II. In this work Dharmakirti is
trying to put (secular) epistemology, the pramanavada, in
the context of Buddhist religion and spirituality. This
explains very well the 'literary masses written as commen-
tary to1 and the 'attitude toward1 PS 1.1ab; in fact the
really important Pramanavarttika II could be regarded
as the starting-point of the blending of epistemology and
Buddhist spirituality, rather than the statement in PS
I.1ab itself.
We will remember that Dignaga regarded the statement by an
expert, aptavada, as a kind of inference in PS II.5ab:
aptavadavisaitivadasamanyad anumanata /
'The statement by an expert is [to be regarded as] in-
ference in so far as there is similarity in trustworthi-
ness [between real inference and reliable statement]'. In
accordance with this thesis, Dignaga could interpret the
words of the Buddha as a kind of valid inference; this is
probably the reason why he did not, in the PS, further
discuss the epistemological problems with regard to the
doctrines taught by the Buddha. The notion that reliable
statements — in so far as they are trustworthy — are a
kind of inference, seems to be adumbrated in the following
passage from NBh:
tadasrayav anumanagamau. tasmin sati syatam
anumanagamav asati ca na syatam
(NBh, p. 10.4-5), 'Inference and tradition [consisting of
reliable statements] are [both] based on a [generally-
accepted fact, drstanta]. When that is present, inference
and [reliable] tradition [as pramanas] are possible; when
that is absent, [ inference and tradition] are
impossible*. The point here in the NBh is that the
validity of inference and reliable tradition / reliable
172
statement is necessarily based on a fact perceived by all
to be true. If both inference and tradition are only valid
through a sound drstanta, it follows that both are similar
in that respect. There is yet another NBh-passage in which
it is said that the validity of an expert statement can be
inferred when it is possible to infer the existence of the
object the statement refers to and when finally the
described object is directly perceived by the hearer of
the statement.
asty atma ity aptopadesat pratiyate, tatranumanam
icchadvesaprayatnasukhaduhkhajnanany atmano lingam iti,
pratyaksam yunjanasya yogasamadhijam atmamanasoh
samyogavisesad atma pratyaksa iti
(NBh, p. 25.1-4), lnThe soul [really] exists11, this is
known from the [reliable] teaching of an expert. With
regard to [the existence of this object and the validity
of] this [teaching about it], there is [the following]
inference: "Desire, aversion, effort, joy, grief and
knowledge are the mark [ie the traits in a person through
which we can infer the existence] of a soul [which has
these properties, in that person] (NS 1.1.10)". The direct
perception [of the soul] for him who exercises himself [in
yoga] is brought forth by contemplation [taught] in a
Yoga[-system]; the soul is directly perceivable through [a
yoga-contemplation consisting in] a special union of the
soul with the mind1. This passage teaches that a
metaphysical object (like the soul), at present not per-
ceived by the hearer, is known to him at first through a
statement by an expert. After that, the hearer infers the
existence of the object the statement has referred to, and
by inferring this, he has also inferred the reliability
and the truth of the statement. If the hearer makes cor-
rect efforts to exercise contemplation (samadhi), he can
direclty perceive the object of the statement. This would
be the ultimate proof of the trustworthiness of the
statement. Thus, Paksilasvamin seems to imply that state-
ments on metaphysical objects can be and must be subjected
to rational inquiry before they are accepted as pramana.
The next logical step would be to say that ultimately
173
there are only two pramanas, perception and inference, and
to subsume reliable statement under inference, as Dignaga
has done.
8. Vetter 1964, p. 31.: 'Keinesfalls soil das uberlieferte
Buddhawort als solches schon AutoritMt sein, nur weil es
jemand gesagt hat, der ftir heilig gilt, und es von vielen
ftlr massgebliche Oberlieferung (agamah) gehalten wird.
Erst wenn gezeigt ist: diese Worte stammen von jemand, der
durch seinen Wandel beweist, das er nicht lttgt und etwas
zu sagen hat, was nicht jedermanns Besitz ist, kann er als
Erkenntnismittel angenommen werden. Das hBchste Ziel des
Handelns muss von einer solchen AutoritSt gegeben werden.
Es ist nSmlich nicht unmittelbar gegeben, sonst wMre es
nicht das h8chste Ziel1.
9. That Dharmaklrti regards tradition, agama, as a statement
by an expert, aptavada, is evident from PVSV, p. 108.1
where he quotes PS II.5ab in which the term aptavada
occurs, and says: ity agamasyanumanatvam uktam (PVSV, p.
108.1-2), 'thus [Dignaga] has said that tradition is
inference1•
10. According to Frauwallner, Devendrabuddhi "is said to have
been a personal pupil of Dharmaklrti; a fact which is in
itself worthy of credence. He must, therefore, have lived
about 630-690 A.D." (Frauwallner 1961, p. 145) ; or
earlier, if we believe Lindtner 1980.
11. My method of interpreting is as follows: I try to under-
stand PV II.1-7 by examining earlier texts that are im-
plicitly or explicitly referred to in it, and which have
formed the basis of the theories propounded in this text.
Since the NBh, the PS and PV I preceded PV II, it is, to
my mind, a sound method to understand theories in PV II as
having grown out of these earlier works, especially in so
far as these earlier texts have formulated a particular
theory for the first time. In order not to neglect the
authoritative explanations given in later times, I have
174
made use of the oldest commentary, the PVP by
Devendrabuddhi, who was a direct pupil of Dharmaklrti.
Therefore, it is very likely that the PVP is closest to
Dharmaklrti's own intentions, rather than later commen-
taries such as, for instance, Prajnakaragupta1s PVBh. This
interpretative translation and explanation of PV II.1-7
are, above all, intended to investigate and make clear
what Dharmaklrti himself wished to say, not what later
commentators thought he said, or should have said.
12. The term avisamvadin is a derivative of avisamvada. Al-
ready in Pali the term avisamvada and variants like
avisamvadaka and avisamvadana occur, cf The Pali Text
Society's Pali-English Dictionary under these entries and
the entries of this term without the privative a_-. For
avisamvadaka and variants it gives the following meanings,
"not deceiving", "not lying", 'honest", "faithful", while
the Critical Pali Dictionary vol I gives for this "not
deceiving", "not breaking one's word", "not false". The
PTS Pali Dictionary gives for visamvada etc, "deceiving",
"untrustworthy", "disappointing", the Critical Pali Dic-
tionary vol I, p. 477 gives "untrustworthiness". Vetter
interprets the word avisamvada as "sich bewMhren" (Vetter
1964, p. 27) and Katsura as "non-contradictoriness"
(Katsura 1984, p. 219). Taking into consideration the
meanings of avisamvada in Pali and the interpretations of
this term in relation to Dharmaklrti's epistemology and
ontology by Vetter and Katsura, it seems to me that
"trustworthiness" renders the various aspects of the term
quite well in English.
13. In PV Il.lab Dharmaklrti equates the means of valid
cognition, pramana, with the resulting cognition, valid
knowledge, jnana. In this, he follows Dignaga who has
stated in PS I.8cd:
savyaparapratitatvat pramanam phalam eva sat //
'[We say that] a means of valid cognition is the resultant
[cognition and not different from it] because [the resul-
tant cognition is] cognized together with the activity [of
175
cognizing the thing through means of valid cognition]'.
For more details about this tenet, consult the chapter on
Dignaga, p. 58 ff.
14. Frauwallner thinks that Sakyamati could have been a direct
pupil of Devendrabuddhi; Sakyamati lived ca 660-720,
Frauwallner 1961, p. 145.
15. My rendering of this term follows Hattori 1968, p. 80. For
a fuller discussion of the interpretation and translationo f arthakriya consult Vetter 1964, p. 13; Nagatomi 1967/68
and Steinkellner 1971, p. 182. Vetter proposes to inter-
pret it as "das ErfMien (kriya) eines Zweck (artha)". He
says '... der Gegenstand alles Erkennens, auch mittelbar
der Schlussfolgerung, [ist] ein Individuel1es
(svalaksanam), nach Raum und Zeit Einmaliges und
Unteilbares, das fShig ist, einen Zweck zu erfilllen
(arthakriyasamartham)', Vetter 1964, p. 13. Nagatomi,
however, criticizes various interpretations, among them
Vetter1s, of arthakriya (in Nagatomi 1967/68, p. 53-54)
for not being satisfactory in all contexts. He thinks
arthakriya has two distinct meanings, first an ontological
one: padarthasya kriya, "the action of a thing, its causal
efficiency", and second an epistemological meaning:
prayoianabhuta kriyaf "action that serves a purpose" (op
cit, p. 53-54, 72), I fail to see a significant difference
between the ontological and the epistemological meanings,
since in both cases arthakriya is a fact, a power predi-
cated of a thing, not an aspect of the knowledge of the
thing. Consequently, arthakriya would only be an ontologi-
cal notion, not an epistemological notion. Steinkellner
sums up the problem as follows: 'Die Dinge "erfilllen einen
Zweck" durch ihr Wirken (kriya), sofern der Mensch eine
solche Zweckerf tlllung von ihnen erwartet (arthah); sie
(arthah) erfflllen einen Zweck (kriya) aber auch -- im
Sinne ihres Zugewandtseins auf ein Ziel — unabhSngig von
dieser Erwartung des Menschen. Nimmt man also die kleine
Undeutlichkeit in Kauf, dass "Zweck" im rein ontologischen
Kontext den Sinn von "Ziel" hat, scheint es wohl am
176
besten, ftlr das Deutsche bei . •, "Erfullen eines Zweckes"
zu bleiben1 , (Steinkellner 1971, p. 182). To me it seems
that the English "production" (kriya) of an "effect"
(artha) is as neutral as the German interpretation of Vet-
ter discussed by Steinkellner in the above quotation.
16. Cf Vetter's interpretation of PV II.1-2 in the following
extremely compressed passage: 'Erkenntnismittel ist
Wissen, das sich in (wir ktinnen ergSnzen: auf die ErliJsung
gerichtetem) Handeln bewShrt. Auch wenn dieses Wissen
durch das Wort Ubermittelt wird, ist es doch
Erkenntnismittel, da man erkennt, was der Sprecher meint,
und was er als Gegenstand seines Handelns hat, auch wenn
einem die Sache selbst nicht unmittelbar zugSnglich ist
(Vetter 1964, p. 32)'.
17. Vetter 1964, p. 59-60, 'Begriff ist ein Wort, wenn es im
Satz steht. Satz ist ein Hinweis im Zusammenhang des
Handelns. wMhrend die Dinge beim einzelnen eine Vorstel-
lung auslttsen, die ihre Beurteilung enthSlt, versucht man,
wenn ein anderer Mensch da ist, durch Worte in diesem
Vorstellungen zu erregen, die ihn auf das entsprechende
Ding ftthren. Dazu mUssen zuerst Wflrter vereinbart werden.
Sie sind nicht von Natur mit den Vorstellungen oder gar
Dingen verknllpft.. • VerstSndigungsmittel ist das Wort also
nur unter Menschen, die eine gleiche Bedeutung mit seinen
Klang verbinden'.
18. Cf Steinkellner 1971, p. 198, note 66: 'Die Lesart mthun-
dnos (Skt sabhava, "having the same nature"] ist ...
sinngemSss richtiger. Eine Konjektur gegen alle
Oberlieferung und die anderen tibetischen Versionen mflchte
ich ... nicht vorschlagen. Dennoch bleibt nicht
einsichtig, warum Dharmakirti an dieser Stelle svabhava-
schreibt1. It is, however, possible to interpret svabhava
here as "being" (bhava) belonging to their own (sva)
[being]", hence "sharing in similar being", "having the
same nature", which semantically would be similar to
sabhava. We must keep in mind that the use of homonyms is
177
a confusing peculiarity of Dharmaklrti's style of writing,
cf for instance his playing with words such as bhava,
arthay bheda/ bhinnay hetu etc in PV I.
19. This sound epistemological principle is already formulated
in NS 1.1.5 in connexion with correct inference when it
says: atha tatpurvakaiti trividham anumanam, 'And preceded
by this [ie by direct perception as pramana] there is
threefold inference1. Here it is maintained that valid in-
ference is generated by valid perception. When this prin-
ciple is applied to Dharmaklrti's epistemology, which ac-
cepts only perception and inference as pramanas, we could
say that valid inferences are drawn only on the basis of
correct and valid direct perceptions of things, and
furthermore, that a valid reliable statement, aptavada,
must be verified by valid perception and inference. Thus,
the ultimate source of knowledge is direct perception. Cf
also NBh, p. 25.1-4 quoted in note 7.
20. Prof. E. Steinkellner in Vienna called my attention to a
passage in Dignaga's Pramanasamuccaya which seems to imply
the same principle, viz that a pramana must make known
hitherto unknown facts. In the commentary on PS 1.3b we
read;
dper na dran pa dan "dod pa dan ze sdan la sogs pa snar
rtogs pa^i don la tshad ma gzan ma yin pa bzin no //
(Hattori 1968, p. 176.25-26), "... Like [various instances
of knowledge] such as recollection, desire, repugnance etc
[which operate] with regard to an object that has pre-
viously been cognized [by direct perception] are not dif-
ferent [ie separate] means of valid cognition1. The im-
plication here seems to be that the first instance in
which an object is cognized, is an instance in which a
previously unknown object is cognized. This constitutes
the direct perception of it, whereas recollection of an
object or desire for and repugnance to an object are sub-
sequent mental reflections, but not means of valid
cognition, precisely because they do not cognize any new
aspects of an object.
178
21. Katsura 1984, an interesting article investigating PV
II.1-6, mentions the two pramana-definitions and notices
that direct perception alone generates new knowledge of
particulars. The two pramana-definitions are called by
Katsura 'the pragmatic criterion of true knowledge1
whereas the fact that perception alone can reveal hitherto
unknown particulars is called by him 'the purely epis-
temological criterion of true knowledge1 (Katsura 1984, p.
214-215). On p. 230 he states his conclusions: '(1) from a
pragmatic point of view ... both perception ... and in-
ference (including verbal knowledge) are non-contradictory
and true in the sense that they can lead to the fulfill-
ment of a human purpose and give new information about the
object. (2) From a purely epistemological point of view ..
only perception is non-erroneous and true in the sense
that it alone possesses the true representation of its
object1. This distinction between pragmatic and purely
epistemological does not seem to have been made by
Dharmaklrti himself in this way, nor by Devendrabuddhi. If
it is to be maintained, we should say that the definition
in PV II. 1 and the discussion on it in PV II.2-5b is the
pragmatic one and the definition offered in PV II.5c is a
purely epistemological one. This is done by Prajnaka-
ragupta in PVBh, p. 30.22 (referred to in Katsura 1984, p.
232, note 28):
tatra paramarthikapramanalaksanam etat,
purvam tu samvyaharikasya,
'With regard to the [passage PV II.5c], this is a defini-
tion of a means of valid cognition related with what
[exists] in the surpreme sense, but the previous
[definition given in PV II.1] is [one] of [a means of
valid cognition as] related with practical [results]1. But
since epistemology, according to Dharmaklrti's own theory,
has got value only in so far as it leads to the effect-
producing particular, that is, in so far as it has prag-
matic value, we need not make this distinction, as it adds
nothing to our understanding of Dharmaklrti's thoughts
about epistemology.
179
22. The notion that knowledge of illusory objects is not
pramana is already expressed in the pratyaksa-definition
in NS 1.1.4; knowledge arising from the contact of the
senses with an object is a means of valid cognition in so
far as this knowledge is avyabhicarin, non-erroneous. Cf
the first chapter which deals with the old Nyaya.
23. More details about anupalabdhi are found in the third
chapter of this book.
24. Grammatically, prapta means "has been acquired", but Prof
Steinkellner in Vienna suggested that I should interpret
it as indicating an unacceptable consequence. Thus prapta
would mean something like prasajyatey "it [wrongly] fol-
lows that". Even though this interpretation of the perfect
passive participle prapta seems somewhat forced, it agrees
with Devendrabuddhi's commentary.
25. The term nihsreyasa is used in NS 1.1.1 to designate the
ultimate goal of Nyaya-philosophy, namely the highest
good, defined in NS 1.1.22 as complete freedom from
suffering, tadatyantavimoksa.
26. Already in the Nyaya-Sutra, we have come across the divi-
sion of reliable statements into those that deal with
visible objects and that describe invisible objects:
sa dvividho drstadrstarthatvat
(NS 1.1.8), fIt [reliable statement] is of two kinds ac-
cording to whether the object [it refers to] is
[something] manifest [to the senses] or not manifest [to
the senses]'.
27. In the translation I follow the Sanskrit which seems to be
the original text on which the Tibetan translation was
based. I owe the Sanskrit fragment to Prof. Steinkellner
in Vienna.
28. We will remember from the chapter on Dharmaklrti's logic
that the svabhava which functions as probans in a
180
svabhavahetu - inference/syllogism must be the same
property (with a wider extension) as the property that is
to be proved. For example, this thing is a simsapa-tree
because it is a tree in general. Simsapa-ness as probandum
has a smaller extension than the probans "treeness in
general"•
29. That the compassion of the Buddha is great is expressed in
PV II.198cd trsl Vetter, p. 99:
tisthanty eva paradhina yesam tu mahatl krpa //
'Diejenigen aber, die [weil sie dies frtiher schon getlbt
hatten] grosses Mitleid haben, leben in der Tat urn anderer
willen [auch nach dem vom Karma festgelegten Lebensende]
weiter1.
30. Cf PV 11.136, trsl Vetter, p. 44:
bahuso bahudhopayam kalena bahunasya ca /
gacchanty abhyasyatas tatra gunadosah prakasatam //fIhm, der intensiv (bahuso) und tlber eine lange Zeit hin
(kalena bahuna) auf vielerlei Weise (bahudha) ein Mittel
tlbt [urn die aus Oberlieferung und Nachdenken gewonnene
Einsicht zu verwirklichen und dann diese und die zu ihrer
Verwirklichung einsetzbaren Mittel anderen zu lehren],
gelangen diesbeztiglich [d.h. beztlglich der verschiedenen
Mittel] die Vor- und Nachteile zu [voller] Klarheit1.
181
Index
A
AKBh
163.1-2 57
334.3-4 48, 61
349.11-12 56
349.12-13 56
Abhidharmakos"a IV, verse
2cd 57
Abhipraya 132
Activity 4, 5
Adhigama 2, 146
Adhigati 154
Adhyaropa 87, 88
AdrstSrtha 31• • •Affirmation 104
Xgama 74, 118, 119, 150,
151, 167, 173
Aksapada Gautama 1, 35
Analogy 41
Anumana 10, 50, 122
Anumeya 45, 51
Anupalabdhi 103, 114
Anupalabdhi-hetu 153
Anvaya 39, 98, 107
Application 27
Apoha 134, 135, 136,
154
Spta 17, 161, 167
Aptavada 118, 122,
171, 173, 177
Aristotelian logic 89
Aristotelian syllogism
44, 89
Aristotelian syllogistic
90
Artha 129, 152
Arthakriya" 126, 128,
129, 130, 150,
152, 155, 175
Arthakriy5sthiti 159
Asapaksa 74, 75, 76,
77, 79, 97
Assemblage 100, 101
Authoritative 115
Authoritativeness 164
Authority 31, 32, 44,
161, 162, 164,
166
Avinabh^va 83, 84
Avinabhavin 52, 101
Avi sarhvada 118, 122,
174
Avisamvadayat i 158
Avisamva:din 1 2 1 , 174
B
Beschaffenheit 112
Beschaffenheitstrager
112
Bhrantijn3na 61, 87
Buddha 115, 116, 117,
119, 124, 125,
143, 157, 158,
160, 161, 162,
163, 164, 165,
166, 167, 168,
169, 171, 180
Buddhism 37
Buddhist epistemological
tradition 170
Buddhist epistemology
170
182
Cause 12, 99, 100,
105, 108, 169
Class 70, 74, 90, 98
Classes 135, 136
Cognition 38, 146, 147,
152, 153, 154
Coldness 105
Commonly acknowledged
fact 22
Comparison 6, 14, 15,
50
Compassion 32, 44, 165,
166, 180
Concept 131, 140, 141
Concepts 86, 132, 134,
136, 138, 141, 142
Conceptual 46, 132
Conceptual ascertainments
137
Conceptual image 1 46
Conceptual knowledge
137, 141, 142,
145, 155, 156
Conceptual superimposition
88
Conceptualization 48,
61, 62, 87, 88
Conceptualizing 58
Conceptualizing mind
134, 147, 148
Concomitant 68
Constancy 126
Contrary 106
Contrary example 27, 54
Contrary examples 43
Contrary instances 65
Conventional 46
Correct knowledge 4
Criterion of truth 157
Deduction 50, 67
Deductive reasoning 71
Denial 104
Denotative words 131,
136
Devendrabuddhi 173
Dharma 17, 32, 166
DharmakTrti 36, 38,
39, 40, 93
Dharman 32
Dhi 141, 142
Dignaga 38, 39, 40
Dpe 79
Drstanta 22, 23, 24,
26, 28, 30, 42,
43, 50, 53, 71,
78, 79, 171, 172
DrstSrtha 31
E
Effect 12, 99, 100,
101, 105, 106,
108, 126, 127,
129, 169
Effect-production 141,
155, 157
Enclosure 139
Erroneous 48
Erroneous knowledge 46
Erroneous perception 60
Essence 98, 99, 104,
105, 107, 108,
,112, 136
Essential property 98,
99, 101, 107,
108, 112
183
Example 22, 23, 27
Exclusion 134, 135
Expert 16, 17, 31, 81
122, 124, 125,
160, 161, 165,
171
Expert statements 118
Extension 68, 70, 98,
101, 102, 105, 106
False perception 61, 87
Fire 52, 67, 69
98, 100, 101,
105, 106, 109,
127, 129
Four noble truths 124,
145, 159
Frauwallner 1933, p 301,
fragm 8 54
Frauwallner 1933, p 480,
fragm 7 52
fragm 7a 53
Frauwallner 1957, p 135,
fragm 1 51
fragm 3 51
fragm 4 52
Frauwallner 1957, p 137,
fragm 9, lines 2-5
46
fragm 9 lines 5-15
48
Frauwallner 1957, p 138,
fragm 9 45
fragm 10 49
fragm 10, line 3 49
Frauwallner 1957, p 36,
fragm 5 53
Gati 147
General definition 120
Generally-accepted fact
23, 27, 53
108
H
Hattori 1968
183.18-19 59
Hetu 22, 52, 74, 76,
89, 94, 95 153,
169
Heya 143, 144, 145
Heyopadeya 143, 144
Highest good 2, 158
Implication 109
Inference 6, 10 14,
29, 40, 49, 56,
57, 62, 64, 73,
94, 118, 123,
151, 152, 153,
159, 164, 171,
172, 177, 180
for oneself 64
for the sake of others
64, 71
Inferendum 49, 51, 65,
66, 67, 71, 72
Inferens 49, 65, 67,
94, 95
Inseparable connexion 67
Intentional effort 77
Intentions 132
184
Invariable concomitance
39, 54, 66, 67, 79,
83, 90, 94, 96, 98,
102, 105, 107, 109,
110, 164
Invariable concomitant
51
Invariably concomitant
49, 52, 70
95, 98, 99,
101, 106, 108,
127, 128
Invariable concomitant
effect 164
J
Jnana 174
K
Kalpana 87, 88
Karya 99, 107, 129
Karyahetu 101, 129
Karyakaranabhava 105
Karyasambhava 129
Kitagawa 1965
447.2 64
455.11-12 65
461.5-9 66
472, text V.4 72
74 64
text V, p 455.1-4
text V, p 472.4-8
Krtaka 77
81
73
Liberation
Linga
Lingin
11/
94,
11
264,
95
74,
Linguistic convention
131, 132
Logic 39, 44, 89
Logical reason 21
Logische Nexus 113
Lun 74
M
MeaningfulMeaningful
Meaningful
Meaningful
Meaningful
160
Means 164,
133sentence
sentences
statement
135
142
133
statements
165
Means of cognition
142
120,
Means of valid cognition
3,
48,
117,
142,
153,
163,
174,
Mental 60
Momentary
N
NB 1.1 36
NBh
10.18 6
10.3-4
10.4-5
12.1 20
1 .5 85
2.1 85
2.2 144
20.3-5
6, 31,
55, 11
121 , 1
148, 1
156, 1
165, 1
177
57
25
171
3
4,
33,
51,
58,
69,
185
223.4-5 166
223.5-224.3 166
223.5 32
223.5-7 32, 81
223.7-8 161
224.5-6 34
225.1-2 19, 32
228.2-4 83
229.5-6 83
24.2 6, 15
24.4 6
24.5 38
25.1-4 172, 177
25.8-26.2 33
28.1-29.1 8
3.1 144
3.1-2 5
3.5 63
3.6 63
30.1-3 9, 47, 61
30.5-7 10
33.3 10, 49
33.4-5 11
34.1 12
34.2 12
34.3-5 12
34.6-35.1 13
35.1-2 13
36.2-4 11
37.1 14
37.1-3 14
37.6-7 15
4.2-4 114
40.1-3 17, 32
41.1-2 18
41.2-4 18
6.1-2 144
6 . 1 - 2 ? p 2 . 2 ; p 3 . 1
1 4 5
NS
73.2-3 26
8.1-3 5
8.3-4 21
81.7 21
81.8 21
82.2-4 22
83.2 23
83.2-3 23
83.5 24
84.1 25
84.1-2 84
85.1-2 26
86.1-3 27
86.11 39
87.3-4 28
87.5-6 28
88.3-4 29
91.3 21
1.1.1 2 , 36, 37,
179
1.1.10 39, 172
1.1.15 39
1.1.1-2 5
1.1.2 3
1.1.22 179
1.1.25 25, 26
1.1.3 6
1.1.32 21
1.1.33 72
1.1.33-39 91
1.1.34 23, 40, 5
1.1.34-35 65
1.1.34-37 52
1.1.35 23, 26
1.1.36-37 52
80
1.1.37 27
1.1.38 27
186
1.1.4 7, 46
1.1.41 20
1.1.5 10, 1
48, 177
1 .1 .6 15
1.1.69 40
1 .1 .7 16
1.1.8 18, 1
1.1.9 37
1.2.1 20
2.1.69 31
2.2.1 83
2.2.2 83
NV 131 .17 111
Nantariyaka 49
Nantariyakatva
Negation 104,
106, 107
Nihsreyasa 31,
Niscaya-manas 1
Non-perception
103, 104,
107, 114
Nyaya 5, 20,
70
Nyayamukha 85
P
PS
1.1 117, 11
1.1a 161
I.1ab 157,
1.2 80
I.2ac 117
1.3b 177
1.3c 58, 61
1.5 59
1.6 60
I.6bc 86
, 48
2, 40,
79
83
105,
158, 179
37
102,
106,
35, 37,
9
170, 171
I.7cd-8ab 60
I.8ab 62
I.8ab, comm line 10-11
I.8cd 62, 63, 142,
174
1.9 148
II 64
II.5ab 118
II.5ab 122, 123,
124, 171, 173
II.5cd 95
II.5d 95
11.11 66
11.18 67
11.20 68, 69, 70
11.25 69
II.5ab 81
II.5cd 65, 74
II.6cd 65
111.9 95
111.10 72, 79
I l l . lab 71
Il l . icd 71
III.2 72
III.9 72, 74
IV.1 78
IV.2ab 79, 96
IV.2b 95
PS 1.1
comm line 1-2 115
comm line 1-7 117
comm line 2-3 115
comm line 3-4 116
comm line 4-5 116
comm line 5-6 116
comm line 6-7 117
187
PV
1.1 94, 97
1.1ab 115
1.166 128
I.166d 129
I.2ac 99
1.2-3 97
1.27 108
1.28 108
1.214 118, 122, 123
1.215 123
1.215-217 121
1 . 2 1 6 124
1 . 2 1 6 - 2 1 7 122
1 . 2 1 6 a - b 118
1 . 2 1 7 1 2 4 , 1 4 3 , 145
1.31 109
1 . 3 6 - 3 7 101
I . 3 a b 102
I . 3 c d 103
1 .4 104
1 . 4 0 - 4 2 135
1 . 4 8 136
1 .49 137
1 . 5 0 - 5 1 137
1 . 6 8 - 7 0 1 3 9 , 1 4 0 ,
146
1 .87 146
1 . 8 7 - 8 8 154
1 . 9 3 143
I I . 1 1 5 7 , 1 6 0 , 178
I I . 1 - 6 1 1 9 , 1 5 7 ,
1 5 8 , 178
I I . 1 - 7 119
I I . 1 3 1 c d - 1 3 2 a b t r s l
V e t t e r , p 39 167
I I . 1 3 2 c d - 1 3 3 a b t r s l
Vetter, p 40 168
II.132cd-133ab, trsl
Vetter, p 40 119
11.136, trsl Vetter, p
44 180
11.145 166
II.145-146a trsl
Vetter, p 52 165
II.198cd trsl Vetter,
p 99 180
Il.lab 122, 125,
151, 158
II.1bc 126, 127
II.1cd 143, 152
II.2 132
II.2-5b 178
II.34a (Mi) 164
II.3ab 138
II.3bd 144, 145
II.4c 147
II.5a 148
II.5c 151, 152,
153, 156, 160,
178
II.5d-6a 154
II.7a 161
II.7cd 162, 163
PV I (Mi) 1-7 120
PV 1.92.93 131
PV II.1 122
PVBh
30.22 178
PVP 120
PVP (P)
2a 1-3 120
2a.4-5 120
2a.5-2b.1 121
2b.2 125
2b.3-5 127
3b.4 130
188
3b.5-7 133
4a.1-2 140
4a.6-7 142
5b.5-6 147
6a.1 148
6a.2-3 148
6b.1-2 151
6b.2-3 152
6b.4 152
6b.5-6 154
6b.8-7a.2 156
7a.4 156
7a.4-5 157
7b.2-5 158
7b.6-7 162
PVSV
108.1 173
108.1-2 173
108.16 123
108.17-109.20 121
108.2-5 119, 150
109.10 118
109.15-16 125r
143, 145
109.15-19 159r 160
109.7-9 118
18.17 114
19.10-12 109
19.8-9 109
2.12-13 97
2.14-17 98
2.16 102
2.19-21 99
2.5-6 96
22.2-3 100
23.18-21 101
3.14 99
3.3-4 99
37.23-27 134
4.14-15 104
4.2-3 102
4.6 102
4.8-11 103
5.9 104
5.9-10 104
6.1-3 105
6.1-6 105
6.3-4 106
6.4 106
6.5 106
6.5-6 106
6.6-7 107
84.4-11 128, 129
PVSVT
392.14-15 123
6.2-3 111
8*24-25 94
8.28-29 95
PVT
Je, p 7a.6-7 94
Je, p 7b.1 95
p 6b.6-7 111
PVT II (P)
p88b.2 126
PVV
p 10, note 1 162
p 11, note 1 163
Paksa 53, 74
Paksilasvamin 36
Paksilasvamin Vatsyayana
1
Paramarthasat 87
Pararthanumana 71, 88,
89, 90, 111
Particular 56, 80, 125,
128, 129, 134,
134, 136, 141,
146
189
Particulars 132, 135,
136, 139, 157
Perception 6, 8, 14,
33, 46, 56, 57,
58, 59, 73, 81
86, 123, 127,
146, 147, 151,
152, 153, 155,
159, 172, 177
Perfection 116
Pervade 68
Pervaded 95, 97
Pervading 97, 106
Pervasion 68, 69, 70,
97
Phala 169
Positive example 28
Positive examples 43
Pramana 3, 4, 7, 37,
38, 55, 63,
115, 117, 118,
119, 123, 124,
126, 134, 142,
145, 147, 148,
150, 151, 153,
154, 155, 157,
158, 163, 164,
167, 169, 172,
174
Pramana-definitions 1 57
Pramana-phala 63
Pramanabhata 11 9
Prama"nata 164
Pramanya 32, 44
Prameyas 36
Pratijna 71
Pratyaksa 7, 50, 58,
60, 81, 127,
146, 147, 153
Prayatnotthita 77
Probandum 21, 22, 23,
24, 51, 98,
107, 108
Probans 22, 24, 50, 52,
71, 95, 98, 99,
102, 105, 107,
108, 114
Produce an effect 129
Produced 77
Product 107
Production 126
Production of an effect
128
Properties 73
Property 11, 14, 21,
52, 64, 65, 66
Property-bearer 11, 14,
72, 73
Reason 21 , 43, 50,
94, 95, 98,
101, 104, 108
Referring 71
Reliability 32
Reliable statement 6,
16, 18, 19, 26,
30, 33, 50, 73,
118, 152, 177
Religion 30
Religious authority 31
Sabda 19, 30, 50,
85, 118, 131
Sadhana 164
Sadharmya 79
Sadharmya-drstanta 79
190
Sadhya 51, 72, 73,
79, 84, 96,
113
Samagrt 100
Samanya 139, 140, 154
Samanyalaksana 56, 80, 85
Sambhava 83, 84
Samketa 132
Samvrta 140, 141
Samvrti 138, 139, 140,
146
SamvrtijnHna 61
Samvrtisaj j fiana 61, 87, 88
Samvrtisat 47, 87
Sapaksa 53, 54, 74,
75, 76, 77, 78,
79, 90, 95, 97,
98
Sastra 148, 149, 150,
152
Secularistic 170
Self-cognition 59
Separation 134, 136
Significative words 138
Similar instances 65
Similarity 154
Simsapa" 98, 99, 102,
107
Smoke 52, 67, 69,
98, 100, 101,
106, 109, 127
128, 129
Statement 122, 132,
145, 172
Sthiti 126
Subject 21, 22, 28,
43, 74
Superimposing 62
Supreme good 3, 31, 144
Sva-samvitti 59
Svabh5va 98, 99, 101,
102, 107, 112,
136, 176, 179
Svabhavahetu 164, 179
Svalaksana 56, 57, 80
85, 128, 132,
175
Svarthanumana 64, 94, 111
Svasamvedana 148
Syllogism 11, 23, 29,
30, 41, 42, 43,
44, 50, 70, 79,
88, 89, 91, 94,
107, 108, 111,
180
Syllogistic proof 108
Tadatmya 105, 113
Thesis 21, 50, 71, 108
Tradition 73, 74, 85,
117, 119, 150,
151, 167, 171,
173
Treatise 148, 149,
150, 151
Treeness 102
Trirupahetu 64
True knowledge 2, 3
Trustworthiness
30, 81, 118,
120, 121, 122,
123, 124, 125,
133, 152, 159,
161, 165, 171,
174
Trustworthy 118, 123,
124, 127, 128
191
U
Uddyotakara 38, 50, 85
Umfassung 112
Uniqueness 56
Universals 56, 57, 139,
157
Upadeya 143f 144, 145
Upalabdhi 38
UpamSna 50, 85
Useful effects 152
Vadavidhana 52, 64,
65, 75
Vadavidhi 36, 45, 50,
58, 61, 66
Vaidharmya 80
Vaidharmya-drsta*nta 79,
80
Validity 148
Vastu 128, 129
Vasubandhu 36, 38, 40,
43, 45, 70
Vedas 31, 34
Vikalpa 134
Vipaksa 53, 54
Viruddha 106
Visaya 71
Visesa 128, 129
Vyapaka 97
Vya"pin 68
VySpta 68
Vyapti 97
VySpya 97, 106
Vyatireka 39, 107, 109
Vyavrtti 136, 154
W
Words 130, 131, 132,
138
Yoga 86, 172
Yogipratyaksa 86, 87
IMPRESSUM
Verleger: Arbeitskreis fur Tibetische und Buddhistische StudienMaria Theresien-StrafSe 3/4/26, 1090 Wien, Austria
Herausgeber und fiir den Inhalt verantwortlich:Prof. Ernst Steinkellner, Reisnerstrafte 6, 1030 Wien
Druck: Ernst Becvar Ges.m.b.H,, Lichtgasse 10, 1150 Wien
WIENER STUDIEN ZUR TIBETOLOGIE UND BUDDHISMUSKUNDE
1: Ernst Steinkellner, Verse-Index of Dharmakirti*s Works (Tibetan Versions).1977. XIV, 225 p. OS 200,-
2: Lobsang Dargyay, Die Legende von den Sieben Prinzessinnen (Saptakumarika-Avadana). In der poetischen Fassung von Guhyadatta/Gopadatta aufgrund dertibetischen Ubersetzung herausgegeben, ubersetzt und bearbeitet. 1978. X, 162p. OS 200,-
3: Piotr Klafkowski, The Secret Deliverance of the Sixth Dalai Lama, as narratedby Dharmatala, Edited from the Hor Chos-' byuri and translated into English,with an introduction and comments.1979. VI, 93 p. OS 200,-
4: Gudrun Buhnemann, Der Allwissende Buddha, Ein Beweis und seine Probleme,Ratnakirti*s Sarvajnasiddhi. 1980. L, 175 p. OS 200,-
5: Helmut Tauscher, Candrakirti - Madhyamakavatarah und Madhyamakavatarabha-?yam (Kapitel VI, Vers 166-226). 1981. XXVII, 214 p. OS 200,-
6: Lobsang Dargyay, Gun than dKon mchog bsTan pa* i sgron me* i rNam thar mdorbsdus bzugs. - A Concise Biography of Gun than dKon mchog bsTan pa* i sgronme. 1981. VI, 45 p. OS 65,-
7: Ernst Steinkellner (Ed.), Gun than dKon mchog bsTan pa*i sgron me*i rNam tharsgo gsum gyi mam bzag pa Legs bsad rgya mtsho*i rba rlabs. 1981. 20 p,OS 35,-
8: Gudrun Buhnemann, Jitari: Kleine Texte, [Description of a manuscript from theBihar Research Society with 10 small texts of Jitari, and the edition of thefollowing texts in Sanskrit: Vedapramanyasiddhi, Sarvajnasiddhi, Nairatmya-siddhi, Jatinirakrti, *Isvaravadimatapariksa.] 1982. 21985. 48 p. OS 100,-
9: Josef KolmaS, Ferdinand Stoliczka (1839-1874): The Life and Work of the CzechExplorer in India and High Asia. 1982, XI, 58 p. OS 80,-
10: E. Steinkellner / H. Tauscher (Ed.), Contributions on Tibetan Language,History and Culture. Proceedings of the Csoma de Koros Symposium held atVelm-Vienna, Austria, 13-19 September 1981, vol.1. 1983. XX, 479 p. OS 560,-
11: E. Steinkellner / H. Tauscher (Ed.), Contributions on Tibetan and BuddhistReligion and Philosophy. Proceedings of the Csoma de Koros Symposium held atVelm-Vienna, Austria, 13-19 September 1981, vol.2. 1983. XII, 334 p. OS 270,-
12: Tilman Vetter, Der Buddha und seine Lehre in Dharmaki rtis Pramanavarttika.Der Abschnitt iiber den Buddha und die vier Edlen Wahrheiten im Pramanasiddhi-Kapitel. Eingeleitet, ediert und ubersetzt. 1984. 1983 p. OS 230,-
13: Andras R6na-Tas, Wiener Vorlesungen zur Sprach- und Kulturgeschichte Tibets.1985. 397 p. OS 490,-
14: Michael Aris, Sburces for the History of Bhutan. 1986, 203 p. OS 250,-
15: Ernst Steinkellner, Dbarmottaras Paralokasiddhi. Nachweis der Wiedergeburt,zugleich eine Widerlegung material istischer Thesen zur Natur der Geistigkeit.Tibetischer Text kritisch herausgegeben und ubersetzt. 1986. 57 p. OS 80,-
16: Per K. Sorensen, Candrakirti - Trisaranasaptati. The Septuagint on the ThreeRefuges. Edited, translated and annotated. 1986. 89 p. OS 120,-
17: David P. Jackson, The Entrance Gate for the Wise (Section III). Sa-skyaPan4ita on Indian and Tibetan Traditions of Pramana and Philosophical Debate.2 vols., 1987. 619 p. OS 620,-
18: Michael Tors ten Much, A Visit to Rahula Sankftyayana* s Collection ofNegatives at the Bihar Research Society: Texts from the Buddhist Epistemo-logical School. 1988. 35 p. OS 52,-
19: Andras R6na-Tas, Mongolisches Lesebuch. Lesestucke in Uigur-MongolischerSchrift mit grammatikalischen Bemerkungen. 1988. 65 p. OS 100,-